# LongBench v2 / 66f3c1ab821e116aacb2ead1

task_id: b6b90ce0-0d61-5ebe-8a42-cef2fbed4f4f
task_key: train--66f3c1ab821e116aacb2ead1
task_revision_id: 3

{"choice_A":"The key themes for France encompass establishing a proactive data policy for big data; focusing on four strategic sectors: healthcare, environment, transport, and defense; enhancing French initiatives in research and development; and preparing for the impact of AI on the workforce.","choice_B":"Federal civilian agencies, excluding those within the DOD or intelligence sectors, allocated USD 973.5 million to AI R&D in FY 2020. This figure increased to USD 1.1 billion after accounting for congressional appropriations and transfers. For FY 2021, these agencies budgeted USD 1.5 billion, which is almost 55% higher than their 2020 request.","choice_C":"As for \"1 + N\" AI program, \"1\" refers to a new generation of major AI science and technology initiatives, concentrating on the forward-looking development of fundamental theories and critical shared technologies. This includes research on big data intelligence, cross-media perception and computing, hybrid enhanced intelligence, collective intelligence, autonomous collaborative control, and decision-making theory. \"N\" pertains to the nationwide planning and implementation of AI research and development projects.","choice_D":"In the fiscal year 2021, non-defense U.S. government agencies allocated a total of $1.53 billion to AI research and development, which is roughly 2.7 times the amount spent in the fiscal year 2019.\nThis amount is expected to increase by 8.8% for the fiscal year 2022, with a total of $1.67 billion requested.","context":"1\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7: \nAI Policy and  \nNational Strategies\nArtificial Intelligence\nIndex Report 2021\n\n\n2\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\nOverview\t\n3\nChapter Highlights\t\n4\n7.1 \u0007\nNATIONAL AND REGIONAL \t\n   \n     AI STRATEGIES\t\n5\nPublished Strategies\t\n6\n\t\n2017\t\n6\n\t\n2018\t\n7\n\t\n2019\t\n9\n\t\n2020\t\n11\nStrategies in Development \t\n   \n(as of December 2020)\t\n12\n\t\nStrategies in Public Consultation\t\n12\nStrategies Announced\t\n13\nHighlight: National AI Strategies  \nand Human Rights\t\n14\n7.2 INTERNATIONAL \n     COLLABORATION ON AI\t\n15\nIntergovernmental Initiatives\t\n15\n\t\nWorking Group\t\n15\n\t\nSummits and Meetings\t\n16\nBilateral Agreements\t\n16\n7.3 U.S. PUBLIC INVESTMENT IN AI\t\n17\nFederal Budget for Non-Defense AI R&D\t 17\nU.S. Department of Defense  \nBudget Request\t\n18\nU.S. Government Contract Spending\t\n19\n\t\nTotal Contract Spending\t\n19\n\t\nContract Spending by  \n\t\nDepartment and Agency\t\n19\n7.4 AI AND POLICYMAKING\t\n21\nLegislation Records on AI\t\n21\n\t\nU.S. Congressional Record\t\n22\n\t\nMentions of AI and ML in  \n\t\nCongressional/Parliamentary \n\t\nProceedings\t\n22\nCentral Banks\t\n24\nU.S. AI Policy Papers\t\n26\nAPPENDIX\t\n27\nChapter Preview\nCHAPTER 7:\nACCESS THE PUBLIC DATA\n\n\n3\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\nOverview\nOVERVIEW\nAI is set to shape global competitiveness over the coming decades, promising \nto grant early adopters a significant economic and strategic advantage. To \ndate, national governments and regional and intergovernmental organizations \nhave raced to put in place AI-targeted policies to maximize the promise of the \ntechnology while also addressing its social and ethical implications. \nThis chapter navigates the landscape of AI policymaking and tracks efforts taking \nplace on the local, national, and international levels to help promote and govern AI \ntechnologies. It begins with an overview of national and regional AI strategies and \nthen reviews activities on the intergovernmental level. The chapter then takes a \ncloser look at public investment in AI in the United States as well as how legislative \nbodies, central banks, and nongovernmental organizations are responding to the \ngrowing need to institute a policy framework for AI technologies.\n\n\n4\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\nCHAPTER\nHIGHLIGHTS\nCHAPTER HIGHLIGHTS\n• \u0007\nSince Canada published the world’s first national AI strategy in 2017, more than 30 other \ncountries and regions have published similar documents as of December 2020.\n• \u0007\nThe launch of the Global Partnership on AI (GPAI) and Organisation for Economic  \nCo-operation and Development (OECD) AI Policy Observatory and Network of Experts  \non AI in 2020 promoted intergovernmental efforts to work together to support the \ndevelopment of AI for all.\n• \u0007\nIn the United States, the 116th Congress was the most AI-focused congressional session in \nhistory. The number of mentions of AI by this Congress in legislation, committee reports, and \nCongressional Research Service (CRS) reports is more than triple that of the 115th Congress. \n\n\n5\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nTo guide and foster the development of AI, countries and regions around the world are establishing strategies and \ninitiatives to coordinate governmental and intergovernmental efforts. Since Canada published the world’s first national \nAI strategy in 2017, more than 30 other countries and regions have published similar documents as of December 2020. \n7.1 NATIONAL AND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nThis section presents an overview of select national and regional AI strategies from around the world, including details on \nthe strategies for G20 countries, Estonia, and Singapore as well as links to strategy documents for many others. Sources \ninclude websites of national or regional governments, the OECD AI Policy Observatory (OECD.AI), and news coverage. “AI \nstrategy” is defined as a policy document that communicates the objective of supporting the development of AI while also \nmaximizing the benefits of AI for society. Excluded are broader innovation or digital strategy documents which do not focus \npredominantly on AI, such as Brazil’s E-Digital Strategy and Japan’s Integrated Innovation Strategy.\nCOUNTRIES  \nWITH PUBLISHED \nAI STRATEGIES: 32\nCOUNTRIES  \nDEVELOPING  \nAI STRATEGIES: 22 \n\n\n6\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nPublished Strategies\n2017\nCanada\n• \u0007\nAI Strategy: Pan Canadian AI Strategy\n• \u0007\nResponsible Organization: Canadian Institute for \nAdvanced Research (CIFAR) \n• \u0007\nHighlights: The Canadian strategy emphasizes \ndeveloping Canada’s future AI workforce, supporting major \nAI innovation hubs and scientific research, and positioning \nthe country as a thought leader in the economic, ethical, \npolicy, and legal implications of artificial intelligence. \n• \u0007\nFunding (December 2020 conversion rate): CAD 125 \nmillion (USD 97 million)\n•  In November 2020, CIFAR published its most recent \nannual report, titled “AICAN,” which tracks progress on \nimplementing its national strategy, which highlighted \nsubstantial growth in Canada’s AI ecosystem, as well \nas research and activities related to healthcare and AI’s \nimpact on society, among other outcomes of the strategy.  \nChina\n• \u0007\nAI Strategy: A Next Generation Artificial Intelligence \nDevelopment Plan\n• \u0007\nResponsible Organization: State Council for the People’s \nRepublic of China\n• \u0007\nHighlights: China’s AI strategy is one of the most \ncomprehensive in the world. It encompasses areas \nincluding R&D and talent development through \neducation and skills acquisition, as well as ethical norms \nand implications for national security. It sets specific \ntargets, including bringing the AI industry in line with \ncompetitors by 2020; becoming the global leader in fields \nsuch as unmanned aerial vehicles (UAVs), voice and \nimage recognition, and others by 2025; and emerging as \nthe primary center for AI innovation by 2030.\n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: China established a New Generation \nAI Innovation and Development Zone in February 2019 \nand released the “Beijing AI Principles” in May 2019 with \na multi-stakeholder coalition consisting of academic \ninstitutions and private-sector players such as Tencent \nand Baidu. \nJapan\n• \u0007\nAI Strategy: Artificial Intelligence Technology Strategy\n• \u0007\nResponsible Organization: Strategic Council for AI \nTechnology\n• \u0007\nHighlights: The strategy lays out three discrete phases of \nAI development. The first phase focuses on the utilization \nof data and AI in related service industries, the second \non the public use of AI and the expansion of service \nindustries, and the third on creating an overarching \necosystem where the various domains are merged. \n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: In 2019, the Integrated Innovation \nStrategy Promotion Council launched another AI strategy, \naimed at taking the next step forward in overcoming \nissues faced by Japan and making use of the country’s \nstrengths to open up future opportunities.\nOthers\nFinland: Finland’s Age of Artificial Intelligence\nUnited Arab Emirates: UAE Strategy for Artificial \nIntelligence\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n7\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nPublished Strategies\n2018\nEuropean Union \n• \u0007\nAI Strategy: Coordinated Plan on Artificial Intelligence\n• \u0007\nResponsible Organization: European Commission \n• \u0007\nHighlights: This strategy document outlines the \ncommitments and actions agreed on by EU member \nstates, Norway, and Switzerland to increase investment \nand build their AI talent pipeline. It emphasizes the value \nof public-private partnerships, creating European data \nspaces, and developing ethics principles.\n• \u0007\nFunding (December 2020 conversation rate): At least \nEUR 1 billion (USD 1.1 billion) per year for AI research and \nat least EUR 4.9 billion (USD 5.4 billion) for other aspects \nof the strategy\n• \u0007\nRecent updates: A first draft of the ethics guidelines was \nreleased in June 2018, followed by an updated version in \nApril 2019.\nFrance\n• \u0007\nAI Strategy: AI for Humanity: French Strategy for Artificial \nIntelligence\n• \u0007\nResponsible Organizations: Ministry for Higher \nEducation, Research and Innovation; Ministry of Economy \nand Finance; Directorate General for Enterprises; Public \nHealth Ministry; Ministry of the Armed Forces; National \nResearch Institute for Digital Sciences; Interministerial \nDirector of the Digital Technology and the Information \nand Communication System \n• \u0007\nHighlights: The main themes include developing \nan aggressive data policy for big data; targeting four \nstrategic sectors, namely health care, environment, \ntransport, and defense; boosting French efforts in \nresearch and development; planning for the impact of AI \non the workforce; and ensuring inclusivity and diversity \nwithin the field.\n• \u0007\nFunding (December 2020 conversion rate): EUR 1.5 \nbillion (USD 1.8 billion) up to 2022\n• \u0007\nRecent Updates: The French National Research Institute \nfor Digital Sciences (Inria) has committed to playing a \ncentral role in coordinating the national AI strategy and \nwill report annually on its progress.\nGermany\n• \u0007\nAI Strategy: AI Made in Germany\n• \u0007\nResponsible Organizations: Federal Ministry of \nEducation and Research; Federal Ministry for Economic \nAffairs and Energy; Federal Ministry of Labour and Social \nAffairs\n• \u0007\nHighlights: The focus of the strategy is on cementing \nGermany as a research powerhouse and strengthening \nthe value of its industries. There is also an emphasis \non the public interest and working to better the lives of \npeople and the environment.\n• \u0007\nFunding (December 2020 conversion rate): EUR 500 \nmillion (USD 608 million) in the 2019 budget and EUR \n3 billion (USD 3.6 billion) for the implementation up to \n2025\n• \u0007\nRecent Updates: In November 2019, the government \npublished an interim progress report on the Germany AI \nstrategy.\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n8\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2018 (continued)\nIndia\n• \u0007\nAI Strategy: National Strategy on  \nArtificial Intelligence: #AIforAll\n• \u0007\nResponsible Organization: National Institution for \nTransforming India (NITI Ayog)\n• \u0007\nHighlights: The Indian strategy focuses on both \neconomic growth and ways to leverage AI to increase \nsocial inclusion, while also promoting research to \naddress important issues such as ethics, bias, and \nprivacy related to AI. The strategy emphasizes sectors \nsuch as agriculture, health, and education, where public \ninvestment and government initiative are necessary.\n• \u0007\nFunding (December 2020 conversion rate): INR 7000 \ncrore (USD 949 million)\n• \u0007\nRecent Updates: In 2019, the Ministry of Electronics and \nInformation Technology released its own proposal to \nset up a national AI program with an allocated INR 400 \ncrore (USD 54 million). The Indian government formed \na committee in late 2019 to push for an organized AI \npolicy and establish the precise functions of government \nagencies to further India’s AI mission.\nMexico\n• \u0007\nAI Strategy: Artificial Intelligence Agenda MX  \n(2019 agenda-in-brief version)\n• \u0007\nResponsible Organization: IA2030Mx, Economía\n• \u0007\nHighlights: As Latin America’s first strategy, the Mexican \nstrategy focuses on developing a strong governance \nframework, mapping the needs of AI in various industries, \nand identifying governmental best practices with an \nemphasis on developing Mexico’s AI leadership.\n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: According to the Inter-American \nDevelopment Bank’s recent fAIr LAC report, Mexico is in \nthe process of establishing concrete AI policies to further \nimplementation.\nUnited Kingdom\n• \u0007\nAI Strategy: Industrial Strategy: Artificial Intelligence \nSector Deal\n• \u0007\nResponsible Organization: Office for Artificial \nIntelligence (OAI)\n• \u0007\nHighlights: The U.K. strategy emphasizes a strong \npartnership between business, academia, and the \ngovernment and identifies five foundations for a \nsuccessful industrial strategy: becoming the world’s most \ninnovative economy, creating jobs and better earnings \npotential, infrastructure upgrades, favorable business \nconditions, and building prosperous communities \nthroughout the country. \n• \u0007\nFunding (December 2020 conversion rate): GBP 950 \nmillion (USD 1.3 billion)\n• \u0007\nRecent Updates: Between 2017 and 2019, the U.K.’s \nSelect Committee on AI released an annual report on the \ncountry’s progress. In November 2020, the government \nannounced a major increase in defense spending of \nGBP 16.5 billion (USD 21.8 billion) over four years, with \na major emphasis on AI technologies that promise to \nrevolutionize warfare.\nOthers\nSweden: National Approach to Artificial Intelligence\nTaiwan: Taiwan AI Action Plan\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n9\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nPublished Strategies\n2019\nEstonia\n• \u0007\nAI Strategy: National AI Strategy 2019–2021\n• \u0007\nResponsible Organization: Ministry of Economic Affairs \nand Communications (MKM)\n• \u0007\nHighlights: The strategy emphasizes actions necessary \nfor both the public and private sectors to take to increase \ninvestment in AI research and development, while also \nimproving the legal environment for AI in Estonia. In \naddition, it hammers out the framework for a steering \ncommittee that will oversee the implementation and \nmonitoring of the strategy.\n• \u0007\nFunding (December 2020 conversion rate): EUR 10 \nmillion (USD 12 million) up to 2021\n• \u0007\nRecent Updates: The Estonian government released an \nupdate on the AI taskforce in May 2019.\nRussia\n• \u0007\nAI Strategy: National Strategy for the Development of \nArtificial Intelligence\n• \u0007\nResponsible Organizations: Ministry of Digital \nDevelopment, Communications and Mass Media; \nGovernment of the Russian Federation\n• \u0007\nHighlights: The Russian AI strategy places a strong \nemphasis on its national interests and lays down \nguidelines for the development of an “information \nsociety” between 2017 and 2030. These include a \nnational technology initiative, departmental projects \nfor federal executive bodies, and programs such as the \nDigital Economy of the Russian Federation, designed to \nimplement the AI framework across sectors. \n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: In December 2020, Russian president \nVladmir Putin took part in the Artificial Intelligence \nJourney Conference, where he presented four ideas for AI \npolicies: establishing experimental legal frameworks for \nthe use of AI, developing practical measures to introduce \nAI algorithms, providing neural network developers with \ncompetitive access to big data, and boosting private \ninvestment in domestic AI industries.\nSingapore\n• \u0007\nAI Strategy: National Artificial Intelligence Strategy\n• \u0007\nResponsible Organization: Smart Nation and Digital \nGovernment Office (SNDGO)\n• \u0007\nHighlights: Launched by Smart Nation Singapore, a \ngovernment agency that seeks to transform Singapore’s \neconomy and usher in a new digital age, the strategy \nidentifies five national AI projects in the following fields: \ntransport and logistics, smart cities and estates, health \ncare, education, and safety and security.\n• \u0007\nFunding (December 2020 conversion rate): While the \n2019 strategy does not mention funding, in 2017 the \ngovernment launched its national program, AI Singapore, \nwith a pledge to invest SGD 150 million (USD 113 million) \nover five years. \n• \u0007\nRecent Updates: In November 2020, SNDGO published \nits inaugural annual update on the Singaporean \ngovernment’s data protection efforts. It describes the \nmeasures taken to date to strengthen public sector data \nsecurity and to safeguard citizens’ private data.\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n10\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2019 (continued)\nUnited States\n• \u0007\nAI Strategy: American AI Initiative\n• \u0007\nResponsible Organization: The White House\n• \u0007\nHighlights: The American AI Initiative prioritizes \nthe need for the federal government to invest in AI \nR&D, reduce barriers to federal resources, and ensure \ntechnical standards for the safe development, testing, \nand deployment of AI technologies. The White House \nalso emphasizes developing an AI-ready workforce and \nsignals a commitment to collaborating with foreign \npartners while promoting U.S. leadership in AI. The \ninitiative, however, lacks specifics on the program’s \ntimeline, whether additional research will be dedicated \nto AI development, and other practical considerations.\n• \u0007\nFunding: N/A \n• \u0007\nRecent Updates: The U.S. government released its \nyear one annual report in February 2020, followed in \nNovember by the first guidance memorandum for federal \nagencies on regulating artificial intelligence applications \nin the private sector, including principles that encourage \nAI innovation and growth and increase public trust and \nconfidence in AI technologies. The National Defense \nAuthorization Act (NDAA) for Fiscal Year 2021 called for a \nNational AI Initiative to coordinate AI research and policy \nacross the federal government.\nSouth Korea\n• \u0007\nAI Strategy: National Strategy for Artificial Intelligence\n• \u0007\nResponsible Organization: Ministry of Science, ICT and \nFuture Planning (MSIP) \n• \u0007\nHighlights: The Korean strategy calls for plans to \nfacilitate the use of AI by businesses and to streamline \nregulations to create a more favorable environment for \nthe development and use of AI and other new industries. \nThe Korean government also plans to leverage its \ndominance in the global supply of memory chips to build \nthe next generation of smart chips by 2030.\n• \u0007\nFunding (December 2020 conversion rate):  \nKRW 2.2 trillion (USD 2 billion)\n• \u0007\nRecent Updates: N/A\nOthers\nColombia: National Policy for Digital Transformation \nand Artificial Intelligence\nCzech Republic: National Artificial Intelligence \nStrategy of the Czech Republic\nLithuania: Lithuanian Artificial Intelligence Strategy: A \nVision for the Future\nLuxembourg: Artificial Intelligence: A Strategic Vision \nfor Luxembourg\nMalta: Malta: The Ultimate AI Launchpad\nNetherlands: Strategic Action Plan for Artificial \nIntelligence\nPortugal: AI Portugal 2030\nQatar: National Artificial Intelligence for Qatar\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n11\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nPublished Strategies\n2020\nIndonesia\n• \u0007\nAI Strategy: National Strategy for the Development of \nArtificial Intelligence (Stranas KA)\n• \u0007\nResponsible Organizations: Ministry of Research \nand Technology (Menristek), National Research and \nInnovation Agency (BRIN), Agency for the Assessment and \nApplication of Technology (BPPT)\n• \u0007\nStrategy Highlights: The Indonesian strategy aims \nto guide the country in developing AI between 2020 \nand 2045. It focuses on education and research, health \nservices, food security, mobility, smart cities, and public \nsector reform.\n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: None\nSaudi Arabia\n• \u0007\nAI Strategy: National Strategy on Data and AI (NSDAI)\n• \u0007\nResponsible Organization: Saudi Data and Artificial \nIntelligence Authority (SDAIA)\n• \u0007\nHighlights: As part of an effort to diversify the country’s \neconomy away from oil and boost the private sector, the \nNSDAI aims to accelerate AI development in five critical \nsectors: health care, mobility, education, government, \nand energy. By 2030, Saudi Arabia intends to train 20,000 \ndata and AI specialists, attract USD 20 billion in foreign \nand local investment, and create an environment that \nwill attract at least 300 AI and data startups. \n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: During the summit where the \nSaudi government released its strategy, the country’s \nNational Center for Artificial Intelligence (NCAI) signed \ncollaboration agreements with China’s Huawei and \nAlibaba Cloud to design AI-related Arabic-language \nsystems.\nOthers\nHungary: Hungary’s Artificial Intelligence Strategy\nNorway: National Strategy for Artificial Intelligence\nSerbia: Strategy for the Development of Artificial \nIntelligence in the Republic of Serbia for the Period \n2020–2025\nSpain: National Artificial Intelligence Strategy\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n12\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nStrategies in Development\n (AS OF DECEMBER 2020)\nStrategies in Public Consultation\nBrazil\n• \u0007\nAI Strategy Draft: Brazilian Artificial Intelligence Strategy\n• \u0007\nResponsible Organization: Ministry of Science, \nTechnology and Innovation (MCTI)\n• \u0007\nHighlights: Brazil’s national AI strategy was announced \nin 2019 and is currently in the public consultation stage. \nAccording to the OECD, the strategy aims to cover \nrelevant topics bearing on AI, including its impact on the \neconomy, ethics, development, education, and jobs, and \nto coordinate specific public policies addressing such \nissues.\n• \u0007\nFunding: N/A\n• \u0007\nRecent Updates: In October 2020, the country’s largest \nresearch facility dedicated to AI was launched in \ncollaboration with IBM, the University of São Paulo, and \nthe São Paulo Research Foundation.\nItaly\n• \u0007\nAI Strategy Draft: Proposal for an Italian Strategy for \nArtificial Intelligence\n• \u0007\nResponsible Organization: Ministry of Economic \nDevelopment (MISE) \n• \u0007\nHighlights: This document provides the proposed \nstrategy for the sustainable development of AI, aimed \nat improving Italy’s competitiveness in AI. It focuses on \nimproving AI-based skills and competencies, fostering AI \nresearch, establishing a regulatory and ethical framework \nto ensure a sustainable ecosystem for AI, and developing \na robust data infrastructure to fuel these developments.\n• \u0007\nFunding (December 2020 conversion rate): EUR 1 \nbillion (USD 1.1 billion) through 2025 and expected \nmatching funds from the private sector, bringing the total \ninvestment to EUR 2 billion.\n• \u0007\nRecent Updates: None\nOthers\nCyprus: National Strategy for Artificial Intelligence\nIreland: National Irish Strategy on Artificial Intelligence\nPoland: Artificial Intelligence Development Policy in \nPoland\nUruguay: Artificial Intelligence Strategy for Digital \nGovernment\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n13\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nStrategies Announced\nArgentina\n• \u0007\nRelated Document: N/A\n• \u0007\nResponsible Organization: Ministry of Science, \nTechnology and Productive Innovation (MINCYT)\n• \u0007\nStatus: Argentina’s AI plan is a part of the Argentine \nDigital Agenda 2030 but has not yet been published. It is \nintended to cover the decade between 2020 and 2030, \nand reports indicate that it has the potential to reap huge \nbenefits for the agricultural sector.\nAustralia\n• \u0007\nRelated Documents: Artificial Intelligence Roadmap /  \nAn AI Action Plan for all Australians\n• \u0007\nResponsible Organizations: Commonwealth Scientific \nand Industrial Research Organisation (CSIRO), Data 61, \nand the Australian government\n• \u0007\nStatus: The Australian government published a road \nmap in 2019 (in collaboration with the national science \nagency, CSIRO) and a discussion paper of an AI action \nplan in 2020 as frameworks to develop a national \nAI strategy. In its 2018–19 budget, the Australian \ngovernment earmarked AUD 29.9 million (USD 22.2 \nmillion [December 2020 conversation rate]) over four \nyears to strengthen the country’s capabilities in AI and \nmachine learning (ML). In addition, CSIRO published a \nresearch paper on Australia’s AI Ethics Framework in 2019 \nand launched a public consultation, which is expected to \nproduce a forthcoming strategy document.\nTurkey\n• \u0007\nRelated Document: N/A\n• \u0007\nResponsible Organizations: Presidency of the Republic \nof Turkey Digital Transformation Office; Ministry of \nIndustry and Technology; Scientific and Technological \nResearch Council of Turkey; Science, Technology and \nInnovation Policies Council\n• \u0007\nStatus: The strategy has been announced but not yet \npublished. According to media sources, it will focus \non talent development, scientific research, ethics and \ninclusion, and digital infrastructure.\nOthers\nAustria: Artificial Intelligence Mission Austria  \n(official report)\nBulgaria: Concept for the Development of Artificial \nIntelligence in Bulgaria Until 2030 (concept document)\nChile: National AI Policy (official announcement)\nIsrael: National AI Plan (news article)\nKenya: Blockchain and Artificial Intelligence Taskforce \n(news article) \nLatvia: On the Development of Artificial Intelligence \nSolutions (official report)\nMalaysia: National Artificial Intelligence (Al) Framework \n(news article)\nNew Zealand: Artificial Intelligence: Shaping a Future \nNew Zealand (official report)\nSri Lanka: Framework for Artificial Intelligence (news \narticle)\nSwitzerland: Artificial Intelligence (official guidelines)\nTunisia: National Artificial Intelligence Strategy (task \nforce announced)\nUkraine: Concept of Artificial Intelligence Development \nin Ukraine AI (concept document)\nVietnam: Artificial Intelligence Development Strategy \n(official announcement)\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n\n\n14\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nRead more on AI national strategies:\n• \u0007\nTim Dutton: An Overview of National AI Strategies\n• \u0007\nOrganisation for Economic Co-operation and Development: OECD AI Policy Observatory\n• \u0007\nCanadian Institute for Advanced Research: Building an AI World, Second Edition\n• \u0007\nInter-American Development Bank: Artificial Intelligence for Social Good in Latin America and the Caribbean:  \nThe Regional Landscape and 12 Country Snapshots\n7.1 NATIONAL \nAND REGIONAL \nAI STRATEGIES\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\nNational AI Strategies and Human Rights\nTable 7.1.1: Mapping human rights \nreferenced in national AI strategies\nHUMAN RIGHTS \nMENTIONED\nSTATES/REGIONAL \nORGANIZATIONS\nThe right to privacy\nAustralia, Belgium, China, \nCzech Republic, Germany, \nIndia, Italy, Luxembourg, Malta, \nNetherlands, Norway, Portugal, \nQatar, South Korea, United \nStates\nThe right \nto equality/\nnondiscrimination\nAustralia, Belgium, Czech \nRepublic, Denmark, Estonia, EU, \nFrance, Germany, Italy, Malta, \nNetherlands, Norway\nThe right to an \neffective remedy\nAustralia (responsibility \nand ability to hold humans \nresponsible), Denmark, Malta, \nNetherlands\nThe rights to \nfreedom of thought, \nexpression, \nand access to \ninformation\nFrance, Netherlands,  \nRussia\nThe right to work\nFrance, Russia\nIn 2020, Global Partners Digital and Stanford’s \nGlobal Digital Policy Incubator published a \nreport examining governments’ national AI \nstrategies from a human rights perspective, \ntitled “National Artificial Intelligence Strategies \nand Human Rights: A Review.” The report \nassesses the extent to which governments \nand regional organizations have incorporated \nhuman rights considerations into their national \nAI strategies and made recommendations to \npolicymakers looking to develop or review AI \nstrategies in the future. \nThe report found that among the 30 states and \ntwo regional strategies (from the European \nUnion and the Nordic-Baltic states), a number \nof strategies refer to the impact of AI on human \nrights, with the right to privacy as the most \ncommonly mentioned, followed by equality \nand nondiscrimination (Table 6.1.1). However, \nvery few strategy documents provide deep \nanalysis or concrete assessment of the impact \nof AI applications on human rights. Specifics \nas to how and the depth to which human \nrights should be protected in the context of \nAI is largely missing, in contrast to the level of \nspecificity on other issues such as economic \ncompetitiveness and innovation advantage. \n\n\n15\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nGiven the scale of the opportunities and the challenges \npresented by AI, a number of international efforts have \nrecently been announced that aim to develop multilateral \nAI strategies. This section provides an overview of those \ninternational initiatives from governments committed to \nworking together to support the development of AI for all. \nThese multilateral initiatives on AI suggest that \norganizations are taking a variety of approaches to \ntackle the practical applications of AI and scale those \nsolutions for maximum global impact. Many countries \nturn to international organizations for global AI norm \nformulation, while others engage in partnerships or \nbilateral agreements. Among the topics under discussion, \nthe ethics of AI—or the ethical challenges raised by current \nand future applications of AI—stands out as a particular \nfocus area for intergovernmental efforts. \nCountries such as Japan, South Korea, the United \nKingdom, the United States, and members of the European \nUnion are active participants of intergovernmental \nefforts on AI. A major AI powerhouse, China, on the other \nhand, has opted to engage in a number of science and \ntechnology bilateral agreements that stress cooperation \non AI as part of the Digital Silk Road under the Belt \nand Road (BRI) initiative framework. For example, AI is \nmentioned in China’s economic cooperation under the BRI \nInitiative with the United Arab Emirates. \nINTERGOVERNMENTAL \nINITIATIVES \nIntergovernmental working groups consist of experts and \npolicymakers from member states who study and report \non the most urgent challenges related to developing and \ndeploying AI and then make recommendations based on \ntheir findings. These groups are instrumental in identifying \nand developing strategies for the most pressing issues in AI \ntechnologies and their applications. \nWorking Groups\nGlobal Partnership on AI (GPAI)\n• \u0007\nParticipants: Australia, Brazil, Canada, France, Germany, \nIndia, Italy, Japan, Mexico, the Netherlands, New \nZealand, South Korea, Poland, Singapore, Slovenia, \nSpain, the United Kingdom, the United States, and the \nEuropean Union (as of December 2020)\n• \u0007\nHost of Secretariat: OECD\n• \u0007\nFocus Areas: Responsible AI; data governance; the future \nof work; innovation and commercialization\n• \u0007\nRecent Activities: Two International Centres of \nExpertise—the International Centre of Expertise in \nMontreal for the Advancement of Artificial Intelligence \nand the French National Institute for Research in Digital \nScience and Technology (INRIA) in Paris—are supporting \nthe work in the four focus areas and held the Montreal \nSummit 2020 in December 2020. Moreover, the data \ngovernance working group published the beta version of \nthe group’s framework in November 2020.\nOECD Network of Experts on AI (ONE AI)\n• \u0007\nParticipants: OECD countries\n• \u0007\nHost: OECD\n• \u0007\nFocus Areas: Classification of AI; implementing \ntrustworthy AI; policies for AI; AI compute\n• \u0007\nRecent Activities: ONE AI convened its first meeting in \nFebruary 2020, when it also launched the OECD AI Policy \nObservatory. In November 2020, the working group on \nthe classification of AI presented the first look at an AI \nclassification framework based on OECD’s definition of AI \ndivided into four dimensions (context, data and input, AI \nmodel, task and output) that aims to guide policymakers \nin designing adequate policies for each type of AI system.\nHigh-Level Expert Group on Artificial Intelligence (HLEG) \n• \u0007\nParticipants: EU countries\n• \u0007\nHost: European Commission\n• \u0007\nFocus Areas: Ethics guidelines for trustworthy AI\n• \u0007\nRecent Activities: Since its launch at the recommendation \n7.2 INTERNATIONAL \nCOLLABORATION ON AI\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.2 INTERNATIONAL \nCOLLABORATION \nON AI\n\n\n16\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nof the EU AI strategy in 2018, HLEG presented the EU Ethics \nGuidelines for Trustworthy Artificial Intelligence and a \nseries of policy and investment recommendations, as \nwell as an assessment checklist related to the guidelines. \nAd Hoc Expert Group (AHEG) for the Recommendation \non the Ethics of Artificial Intelligence\n• \u0007\nParticipants: United Nations Educational, Scientific and \nCultural Organization (UNESCO) member states\n• \u0007\nHost: UNESCO\n• \u0007\nFocus Areas: Ethical issues raised by the development \nand use of AI \n• \u0007\nRecent Activities: The AHEG produced a revised first draft \nRecommendation on the Ethics of Artificial Intelligence, \nwhich was transmitted in September 2020 to Member States \nof UNESCO for their comments by December 31, 2020.\nSummits and Meetings\nAI for Good Global Summit\n• \u0007\nParticipants: Global (with the United Nations and its \nagencies)\n• \u0007\nHosts: International Telecommunication Union, XPRIZE \nFoundation\n• \u0007\nFocus Areas: Trusted, safe, and inclusive development of \nAI technologies and equitable access to their benefits\nAI Partnership for Defense\n• \u0007\nParticipants: Australia, Canada, Denmark, Estonia, \nFinland, France, Israel, Japan, Norway, South Korea, \nSweden, the United Kingdom, and the United States\n• \u0007\nHosts: Joint Artificial Intelligence Center, U.S. \nDepartment of Defense\n• \u0007\nFocus Areas: AI ethical principles for defense\nChina-Association of Southeast Asian Nations (ASEAN) \nAI Summit\n• \u0007\nParticipants: Brunei, Cambodia, China, Indonesia, Laos, \nMalaysia, Myanmar, the Philippines, Singapore, Thailand, \nand Vietnam\n• \u0007\nHosts: China Association for Science and Technology, \nGuangxi Zhuang Autonomous Region, China\n• \u0007\nFocus Areas: Infrastructure construction, digital \neconomy, and innovation-driven development\nBILATERAL AGREEMENTS\nBilateral agreements focusing on AI are another form \nof international collaboration that has been gaining in \npopularity in recent years. AI is usually included in the \nbroader context of collaborating on the development of \ndigital economies, though India stands apart for investing \nin developing multiple bilateral agreements specifically \ngeared toward AI.\nIndia and United Arab Emirates\nInvest India and the UAE Ministry of Artificial Intelligence \nsigned a memorandum of understanding in July 2018 \nto collaborate on fostering innovative AI ecosystems \nand other policy concerns related to AI. Two countries \nwill convene a working committee aimed at increasing \ninvestment in AI startups and research activities in \npartnership with the private sector.\nIndia and Germany\nIt was reported in October 2019 that India and Germany \nlikely will sign an agreement including partnerships on the \nuse of artificial intelligence (especially in farming).\nUnited States and United Kingdom\nThe U.S. and the U.K. announced a declaration in \nSeptember 2020, through the Special Relationship \nEconomic Working Group, that the two countries will \nenter into a bilateral dialogue on advancing AI in line with \nshared democratic values and further cooperation in AI \nR&D efforts.\nIndia and Japan\nIndia and Japan were said to have finalized an agreement \nin October 2020 that focuses on collaborating on digital \ntechnologies, including 5G and AI. \nFrench and Germany\nFrance and Germany signed a road map for a Franco-\nGerman Research and Innovation Network on artificial \nintelligence as part of the Declaration of Toulouse \nin October 2019 to advance European efforts in the \ndevelopment and application of AI, taking into account \nethical guidelines.\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.2 INTERNATIONAL \nCOLLABORATION \nON AI\n\n\n17\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2020 (Request)\n2020 (Enacted)\n2021 (Request)\n0\n500\n1,000\n1,500\nBudget (in Millions of U.S. Dollars)\nU.S. FEDERAL BUDGET for NON-DEFENSE AI R&D, FY 2020-21\nSource: U.S. NITRD Program, 2020 | Chart: 2021 AI Index Report\nFEDERAL BUDGET FOR  \nNON-DEFENSE AI R&D \nIn September 2019, the White House \nNational Science and Technology Council \nreleased a report attempting to total \nup all public-sector AI R&D funding, the \nfirst time such a figure was published. \nThis funding is to be disbursed as grants \nfor government laboratories or research \nuniversities or in the form of government \ncontracts. These federal budget figures, \nhowever, do not include substantial AI \nR&D investments by the Department of \nDefense (DOD) and the intelligence sector, \nas they were withheld from publication for \nnational security reasons. \nAs shown in Figure 7.3.1, federal civilian \nagencies—those agencies that are not part \nof the DOD or the intelligence sector—\nallocated USD 973.5 million to AI R&D \nfor FY 2020, a figure that rose to USD 1.1 \nbillion once congressional appropriations \nand transfers were factored in. For FY \n2021, federal civilian agencies budgeted \nUSD 1.5 billion, which is almost 55% \nhigher than its 2020 request. \n7.3 U.S. PUBLIC INVESTMENT IN AI\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.3 U.S. PUBLIC \nINVESTMENT \nIN AI\nThis section examines public investment in AI in the United States based on data from the U.S. Networking and Information \nTechnology Research and Development (NITRD) program and Bloomberg Government. \nFigure 7.3.1\nFederal civilian agencies—those \nagencies that are not part of the \nDOD or the intelligence sector—\nallocated USD 973.5 million to \nAI R&D for FY 2020, a figure \nthat rose to USD 1.1 billion once \ncongressional appropriations \nand transfers were factored in.\n\n\n18\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2018 (Enacted)\n2019 (Enacted)\n2020 (Enacted)\n2021 (Request)\n0\n1,000\n2,000\n3,000\n4,000\n5,000\nBudget (in Millions of U.S. Dollars)\n927\n841\nDOD Reported\nBudget on AI R&D\nDOD Reported\nBudget on AI R&D\nU.S. DOD BUDGET for AI-SPECIFIC RESEARCH DEVELOPMENT, TEST, and EVALUATION (RDT&E), FY 2018-20\nSources: Bloomberg Government & U.S. Department of Defense, 2020 | Chart: 2021 AI Index Report\nFigure 7.3.2\nU.S. DEPARTMENT OF DEFENSE AI \nR&D BUDGET REQUEST \nWhile the official DOD budget is not publicly available, \nBloomberg Government has analyzed the department’s \npublicly available budget request for research, \ndevelopment, test, and evaluation (RDT&E)— data that \nsheds light on its spending on AI R&D. \nWith 305 unclassified DOD R&D programs specifying the use \nof AI or ML technologies, the combined U.S. military budget \nfor AI R&D in FY 2021 is USD 5.0 billion (Figure 7.3.2). This \nfigure appears consistent with the USD 5.0 billion enacted \nthe previous year. However, the FY 2021 figure reflects \na budget request, rather than a final enacted budget. \nAs noted above, once congressional appropriations are \nfactored in, the true level of funding available to DOD AI R&D \nprograms in FY 2021 may rise substantially.\nThe top five projects set to receive the highest amount of \nAI R&D investment in FY 2021: \n• \u0007\nRapid Capability Development and Maturation, by the \nU.S. Army (USD 284.2 million)\n• \u0007\nCounter WMD Technologies and Capabilities \nDevelopment, by the DOD Threat Reduction Agency \n(USD 265.2 million) \n• \u0007\nAlgorithmic Warfare Cross-Functional Team (Project \nMaven), by the Office of the Secretary of Defense (USD \n250.1 million) \n• \u0007\nJoint Artificial Intelligence Center (JAIC), by the Defense \nInformation Systems Agency (USD 132.1 million)\n• \u0007\nHigh Performance Computing Modernization Program, \nby the U.S. Army (USD 99.6 million)\nIn addition, the Defense Advanced Research Projects \nAgency (DARPA) alone is investing USD 568.4 million in AI \nR&D, an increase of USD 82 million from FY 2020.\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.3 U.S. PUBLIC \nINVESTMENT \nIN AI\nImportant data caveat: This chart illustrates the challenge of working with contemporary government data sources \nto understand spending on AI. By one measure—the requests that include AI-relevant keywords—the DOD is requesting \nmore than USD 5 billion for AI-specific research development in 2021 . However, DOD’s own accounting produces a \nradically smaller number: USD 841 million. This relates to the issue of defining where an AI system ends and another \nsystem begins; for instance, an initiative that uses AI for drones may also count hardware-related expenditures for the \ndrones within its “AI” budget request, though the AI software component will be much smaller.\n\n\n19\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nUSD 1.5 billion agencies spent in FY 2019 (Figure 7.3.3). \nAI spending in 2020 was more than six times higher than \nwhat it was just five years ago—about USD 300 million in \nFY 2015. However, to put this in perspective, the federal \ngovernment spent USD 682 billion on contracts in FY 2020, \nso AI currently represents 0.25% of government spending.\nContract Spending by Department and Agency\nFigure 7.3.4 shows that in FY 2020, the DOD spent more on \nAI-related contracts than any other federal department \nor agency (USD 1.4 billion). In second and third place \nare NASA (USD 139.1 million) and the Department of \nHomeland Security (USD 112.3 million). DOD, NASA, and \nthe Department of Health and Human Services top the \nlist for the most contract spending on AI over the past 10 \nyears combined (Figure 7.3.5). In fact, DOD’s total contract \nspending on AI from 2001 to 2020 (USD 3.9 billion) is more \nthan what was spent by the other 44 departments and \nagencies combined (USD 2.9 billion) over the same period.\nLooking ahead, DOD spending on AI contracts is only \nexpected to grow as the Pentagon’s Joint Artificial \nIntelligence Center (JAIC), established in June 2018, is \n2001\n2002\n2003\n2004\n2005\n2006\n2007\n2008\n2009\n2010\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n0\n500\n1,000\n1,500\n2,000\nContract Spending (in Millions of U.S. Dollars)\n1,837\nU.S. GOVERNMENT TOTAL CONTRACT SPENDING on AI, FY 2001-20\nSource: Bloomberg Government, 2020 | Chart: 2021 AI Index Report\nFigure 7.3.3\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.3 U.S. PUBLIC \nINVESTMENT \nIN AI\nU.S. GOVERNMENT AI-RELATED \nCONTRACT SPENDING\nAnother indicator of public investment in AI technologies is \nthe level of spending on government contracts across the \nfederal government. Contracting for products and services \nsupplied by private businesses typically occupies the largest \nshare of an agency’s budget. Bloomberg Government built \na model that captures contract spending on AI technologies \nby adding up all contracting transactions that contain a \nset of more than 100 AI-specific keywords in their titles or \ndescriptions. The data reveals that the amount the federal \ngovernment spends on contracts for AI products and \nservices has reached an all-time high and shows no sign of \nslowing down. However, note that during the procurement \nprocess, vendors may add a bunch of keywords into their \napplications, so some of these things may have a relatively \nsmall AI component relative to other parts of technology.\nTotal Contract Spending\nFederal departments and agencies spent a combined \nUSD 1.8 billion on unclassified AI-related contracts in FY \n2020. This represents a more than 25% increase from the \n\n\n20\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n0\n500\n1000\n1500\n2000\n2500\n3000\n3500\n4000\nContract Spending (in Millions of U.S. Dollars)\nDepartment of Defense (DOD)\nNational Aeronautics and\nSpace Administration (NASA)\nDepartment of Health and\nHuman Services (HHS)\nDepartment of the Treasury\n(TREAS)\nDepartment of Homeland\nSecurity (DHS)\nDepartment of Veterans A airs\n(VA)\nDepartment of Commerce\n(DOC)\nDepartment of Agriculture\n(USDA)\nGeneral Services\nAdministration (GSA)\nDepartment of State (DOS)\nTOP 10 CONTRACT SPENDING on AI by U.S. GOVERNMENT DEPARTMENT and AGENCY, 2001-20 (SUM)\nSource: Bloomberg Government, 2020 | Chart: 2021 AI Index Report\n0\n200\n400\n600\n800\n1,000\n1,200\n1,400\nContract Spending (in Millions of U.S. Dollars)\nDepartment of Defense (DOD)\nNational Aeronautics and Space\nAdministration (NASA)\nDepartment of Homeland Security\n(DHS)\nDepartment of Health and Human\nServices (HHS)\nDepartment of Commerce (DOC)\nDepartment of the Treasury\n(TREAS)\nDepartment of Veterans A airs\n(VA)\nSecurities and Exchange\nCommission (SEC)\nDepartment of Agriculture (USDA)\nDepartment of Justice (DOJ)\nTOP 10 CONTRACT SPENDING on AI by U.S. GOVERNMENT DEPARTMENT and AGENCY, 2020\nSource: Bloomberg Government, 2020 | Chart: 2021 AI Index Report\nFigure 7.3.4\nFigure 7.3.5\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.3 U.S. PUBLIC \nINVESTMENT \nIN AI\nstill in the early stages of driving DOD’s AI spending. In \n2020, JAIC awarded two massive contracts, one to Booz \nAllen Hamilton for the five-year, USD 800 million Joint \nWarfighter program, and another to Deloitte Consulting for \na four-year, USD 106 million enterprise cloud environment \nfor the JAIC, known as the Joint Common Foundation.\n\n\n21\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n107Th\n(2001-2002)\n108Th\n(2003-2004)\n109Th\n(2005-2006)\n110Th\n(2007-2008)\n111Th\n(2009-2010)\n112Th\n(2011-2012)\n113Th\n(2013-2014)\n114Th\n(2015-2016)\n115Th\n(2017-2018)\n116th\n(2019-2020)\n0\n100\n200\n300\n400\n500\nNumber of Mentions\n486\n149\n22\n10\n16\n15\n17\n4\n8\n7\n243\n173\n44\n66\n39\n70\nMENTIONS of AI in U.S. CONGRESSIONAL RECORD by LEGISLATIVE SESSION, 2001-20\nSource: Bloomberg Government, 2020 | Chart: 2021 AI Index Report\nCongressional Research Service Reports\nCommittee Reports\nLegislation\nAs AI gains attention and importance, policies and \ninitiatives related to the technology are becoming higher \npriorities for governments, private companies, technical \norganizations, and civil society. This section examines \nhow three of these four are setting the agenda for AI \npolicymaking, including the legislative and monetary \nauthority of national governments, as well as think tanks, \ncivil society, and the technology and consultancy industry. \nLEGISLATION RECORDS ON AI\nThe number of congressional and parliamentary \nrecords on AI is an indicator of governmental interest \nin developing AI capabilities—and legislating issues \npertaining to AI. In this section, we use data from \nBloomberg and McKinsey & Company to ascertain the \n7.4 AI AND POLICYMAKING\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\nFigure 7.4.1\nnumber of these records and how that number has \nevolved in the last 10 years. \nBloomberg Government identified all legislation (passed \nor introduced), reports published by congressional \ncommittees, and CRS reports that referenced one or more \nAI-specific keywords. McKinsey & Company searched for \nthe terms “artificial intelligence” and “machine learning” \non the websites of the U.S. Congressional Record, the U.K. \nParliament, and the Parliament of Canada. For the United \nStates, each count indicates that AI or ML was mentioned \nduring a particular event contained in the Congressional \nRecord, including the reading of a bill; for the U.K. and \nCanada, each count indicates that AI or ML was mentioned \nin a particular comment or remark during the proceedings.1\n1 If a speaker or member mentioned artificial intelligence (AI) or machine learning (ML) multiple times within remarks, or multiple speakers mentioned AI or ML within the same event, it appears only \nonce as a result. Counts for AI and ML are separate, as they were conducted in separate searches. Mentions of the abbreviations “AI” or “ML” are not included.\n\n\n22\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n0\n20\n40\n60\n80\n100\n120\n140\nNumber of Mentions\n120\n129\n92\n27\n0\n9\n8\n1\n1\n1\n101\n92\n28\n28\n25\n23\n67\n6\n7\nMENTIONS of AI and ML in the PROCEEDINGS of U.S. CONGRESS, 2011-20\nSources: U.S. Congressional Record website, the McKinsey Global Institute, 2020 | Chart: 2021 AI Index Report\nMachine Learning\nArtificial Intelligence\nU.S. Congressional Record\nThe 116th Congress (January 1, 2019–January 3, 2021) is \nthe most AI-focused congressional session in history. The \nnumber of mentions of AI by this Congress in legislation, \ncommittee reports, and CRS reports is more than triple \nthat of the 115th Congress. Congressional interest in AI \nhas continued to accelerate in 2020. Figure 7.4.1 shows \nthat during this congressional session, 173 distinct \npieces of legislation either focused on or contained \nlanguage about AI technologies, their development, \nuse, and rules governing them. During that two-year \nperiod, various House and Senate committees and \nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\nFigure 7.4.2\nsubcommittees commissioned 70 reports on AI, while \nthe CRS, tasked as a fact-finding body for members of \nCongress, published 243 about AI or referencing AI. \nMentions of AI and ML in Congressional/\nParliamentary Proceedings\nAs shown in Figures 7.4.2–7.4.5, the number of mentions \nof artificial intelligence and machine learning in the \nproceedings of the U.S. Congress and the U.K. parliament \ncontinued to rise in 2020, while there were fewer \nmentions in the parliamentary proceedings of Canada. \n\n\n23\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n0\n50\n100\n150\n200\n250\n300\nNumber of Mentions\n283\n192\n183\n138\n51\n0\n4\n5\n7\n1\n246\n158\n138\n179\n34\n42\n37\nMENTIONS of AI and ML in the PROCEEDINGS of U.K. PARLIAMENT, 2011-20\nSources: Parliament of U.K. website, the McKinsey Global Institute, 2020 | Chart: 2021 AI Index Report\nMachine Learning\nArtificial Intelligence\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n0\n10\n20\n30\n40\nNumber of Mentions\n34\n38\n18\n21\n0\n0\n0\n0\n0\n2\n35\n33\n21\n17\n3\nMENTIONS of AI and ML in the PROCEEDINGS of CANADIAN PARLIAMENT, 2011-20\nSources: Canadian Parliament website, the McKinsey Global Institute, 2020 | Chart: 2021 AI Index Report\nMachine Learning\nArtificial Intelligence\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\nFigure 7.4.3\nFigure 7.4.4\n\n\n24\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\n2 See Science & Technology Review and Scientific American for more details.\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n0\n200\n400\n600\n800\n1,000\nNumber of Mentions\n225\nMENTIONS of AI in CENTRAL BANK COMMUNICATIONS around THE WORLD, 2011-20\nSource: Prattle/LiquidNet, 2020 | Chart: 2021 AI Index Report\nFigure 7.4.5\nCENTRAL BANKS\nCentral banks play a key role in conducting currency and \nmonetary policy in a country or a monetary union. As \nwith many other institutions, central banks are tasked \nwith integrating AI into their operations and relying on \nbig data analytics to assist them with forecasting, risk \nmanagement, and financial supervision. \nPrattle, a leading provider of automated investment \nresearch solutions, monitors mentions of AI in the \ncommunications of central banks, including meeting \nminutes, monetary policy papers, press releases, \nspeeches, and other official publications. \nFigure 7.4.5 shows a significant increase in the mention \nof AI across 16 central banks over the past 10 years, \nwith the number reaching a peak of 1,020 in 2019. The \nsharp decline in 2020 can be explained by the COVID-19 \npandemic as most central bank communications focused \non responses to the economic downturn. Moreover, \nthe Federal Reserve in the United States, Norges Bank \nin Norway, and the European Central Bank top the \nlist for the most aggregated number of AI mentions in \ncommunications in the past five years (Figure 7.4.6). \n\n\n25\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n0\n200\n400\n600\n800\n1,000\n1,200\n1,400\n1,600\n1,800 2,000\nNumber of Mentions\nFederal Reserve\nNorges Bank\nEuropean Central Bank\nReserve Bank of India\nBank of England\nBank of Israel\nBank of Japan\nBank of Korea\nReserve Bank of Australia\nReserve bank of New Zealand\nBank of Taiwan\nBank of Canada\nSveriges Riksbank\nSwedish Riksbank\nCentral Bank of the Republic of Turkey\nCentral Bank of Brazil\nMENTIONS of AI in CENTRAL BANK COMMUNICATIONS around THE WORLD by BANK, 2016-20 (SUM)\nSource: Prattle/LiquidNet, 2020 | Chart: 2021 AI Index Report\nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\nFigure 7.4.6\n\n\n26\nCHAPTER 7 PREVIEW\nArtificial Intelligence\nIndex Report 2021\n0\n20\n40\n60\n80\n100\n120\n140\n160\nNumber of Policy Products\nInnovation & Technology\nInt'l Affairs & Int'l Security\nIndustry & Regulation\nWorkforce & Labor\nGovernment & Public Administration\nPrivacy, Safety & Security\nEthics\nJustice & Law Enforcement\nEquity & Inclusion\nEducation & Skills\nSocial & Behavioral Sciences\nHealth & Biological Sciences\nCommunications & Media\nDemocracy\nHumanities\nEnergy & Environment\nPhysical Sciences\nU.S. AI POLICY PRODUCTS by TOPIC, 2019-20 (SUM)\nSource: Stanford HAI & AI Index, 2020 | Chart: 2021 AI Index Report\nSecondary Topic\nPrimary Topic\nU.S. AI POLICY PAPERS\nWhat are the AI policy initiatives outside national and \nintergovernmental governments? We monitored 42 \nprominent organizations that deliver policy papers on \ntopics related to AI and assessed the primary topic as \nwell as the secondary topic on policy papers published \nin 2019 and 2020. (See the Appendix for a complete list \nof organizations included.) Those organizations are \neither U.S.-based or have a sizable presence in the United \nStates, and we grouped them into three categories: think \ntanks, policy institutes and academia (27); civil society \norganizations, associations and consortiums (9); and \nindustry and consultancy (6). \nAI policy papers are defined as research papers, research \nreports, blog posts, and briefs that focus on a specific policy \nissue related to AI and provide clear recommendations \nCHAPTER 7:\nAI POLICY AND  \nNATIONAL STRATEGIES\n7.4 AI AND \nPOLICYMAKING\nFigure 7.4.7\nfor policymakers. Primary topics mean that such a topic is \nthe main focus of the policy paper, while secondary topics \nmean that the policy paper either briefly touches on the \ntopic or the topic is a sub-focus of the paper. \nCombined data for 2019 and 2020 suggests that the topics \nof innovation and technology, international affairs and \ninternational security, and industry and regulation are \nthe main focuses of AI policy papers in the United States \n(Figure 7.4.7). Fewer documents placed a primary focus \non topics related to AI ethics—such as ethics, equity and \ninclusion; privacy, safety and security; and justice and law \nenforcement—which have largely been secondary topics. \nMoreover, topics bearing on the physical sciences, energy \nand environment, humanities, and democracy have \nreceived the least attention in U.S. AI policy papers. \n\n\n27\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7 PREVIEW\nAPPENDIX\nCHAPTER 7:  \nAI POLICY AND  \nNATIONAL STRATEGIES\nAPPENDIX\nBLOOMBERG GOVERNMENT \nBloomberg Government (BGOV) is a subscription-\nbased market intelligence service designed to make \nU.S. government budget and contracting data more \naccessible to business development and government \naffairs professionals. BGOV’s proprietary tools ingest \nand organize semi-structured government data sets \nand documents, enabling users to track and forecast \ninvestment in key markets. \nMethodology\nThe BGOV data included in this section was drawn from \nthree original sources:\nContract Spending: BGOV’s Contracts Intelligence Tool \ningests on a twice-daily basis all contract spending data \npublished to the beta.SAM.gov Data Bank, and structures \nthe data to ensure a consistent picture of government \nspending over time. For the section “U.S. Government \nContract Spending,” BGOV analysts used FPDS-NG data, \norganized by the Contracts Intelligence Tool, to build a \nmodel of government spending on artificial intelligence-\nrelated contracts in the fiscal years 2000 through 2021. \nBGOV’s model used a combination of government-\ndefined produce service codes and more than 100 \nAI-related keywords and acronyms to identify AI-related \ncontract spending.\nDefense RDT&E Budget: BGOV organized all 7,057 \nbudget line items included in the RDT&E budget request \nbased on data available on the DOD Comptroller website. \nFor the section “U.S. Department of Defense (DOD) \nBudget,” BGOV used a set of more than a dozen AI-\nspecific keywords to identify 305 unique budget activities \nrelated to artificial intelligence and machine learning \nworth a combined USD 5.0 billion in FY 2021.\nCongressional Record (available on Congressional \nRecord website): BGOV maintains a repository of \ncongressional documents, including bills, amendments, \nbill summaries, Congressional Budget Office \nassessments, reports published by congressional \ncommittees, Congressional Research Service (CRS), and \nothers. For the section “U.S. Congressional Record,” \nBGOV analysts identified all legislation (passed or \nintroduced), congressional committee reports, and \nCRS reports that referenced one or more of a dozen AI-\nspecific keywords. Results are organized by a two-year \ncongressional session.\nLIQUIDNET\nPrepared by Jeffrey Banner and Steven Nichols\nSource\nLiquidnet provides sentiment data that predicts \nthe market impact of central bank and corporate \ncommunications. Learn more about Liquidnet here. \nExamples of Central Bank Mentions\nHere are some examples of how AI is mentioned by \ncentral banks: In the first case, China uses a geopolitical \nenvironment simulation and prediction platform \nthat works by crunching huge amounts of data and \nthen providing foreign policy suggestions to Chinese \ndiplomats or the Bank of Japan use of AI prediction \nmodels for foreign exchange rates. For the second \ncase, many central banks are leading communications \nthrough either official documents—for example, on \nJuly 25, 2019, the Dutch Central Bank (DNB) published \nGuidelines for the use of AI in financial services and \nlaunched its six “SAFEST” principles for regulated firms \nto use AI responsibly—or a speech on June 4, 2019, by \nthe Bank of England’s Executive Director of U.K. Deposit \nTakers Supervision James Proudman, titled “Managing \nMachines: The Governance of Artificial Intelligence,” \nfocused on the increasingly important strategic issue of \nhow boards of regulated financial services should use AI. \n\n\n28\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7 PREVIEW\nCHAPTER 7:  \nAI POLICY AND  \nNATIONAL STRATEGIES\nAPPENDIX\nMCKINSEY GLOBAL INSTITUTE\nSource\nData collection and analysis was performed by the \nMcKinsey Global Institute (MGI).\nCanada (House of Commons)\nData was collected using the Hansard search feature on \nParliament of Canada website. MGI searched for the terms \n“Artificial Intelligence” and “Machine Learning” (quotes \nincluded) and downloaded the results as a CSV. The date \nrange was set to “all debates.” Data is as of Dec. 31, 2020. \nData are available online from Aug. 31, 2002.\nEach count indicates that Artificial Intelligence or Machine \nLearning was mentioned in a particular comment or remark \nduring the proceedings of the House of Commons. This \nmeans that within an event or conversation, if a member \nmentions AI or ML multiple times within their remarks, it \nwill appear only once. However if, during the same event, \nthe speaker mentions AI or ML in separate comments (with \nother speakers in between), it will appear multiple times. \nCounts for Artificial Intelligence or Machine Learning are \nseparate, as they were conducted in separate searches. \nMentions of the abbreviations AI or ML are not included. \nUnited Kingdom (House of Commons, House of \nLords, Westminster Hall, and Committees)\nData was collected using the Find References feature of the \nHansard website of the U.K. Parliament. MGI searched for \nthe terms “Artificial Intelligence” and “Machine Learning” \n(quotes included) and catalogued the results. Data is as \nof Dec. 31, 2020. Data are available online from January 1, \n1800 onward. Contains Parliamentary information licensed \nunder the Open Parliament Licence v3.0.\nAs in Canada, each count indicates that Artificial \nIntelligence or Machine Learning was mentioned in a \nparticular comment or remark during a proceeding. \nTherefore, if a member mentions AI or ML multiple times \nwithin their remarks, it will appear only once. However \nif, during the same event, the same speaker mentions \nAI or ML in separate comments (with other speakers in \nbetween), it will appear multiple times. Counts for Artificial \nIntelligence or Machine Learning are separate, as they \nwere conducted in separate searches. Mentions of the \nabbreviations AI or ML are not included. \nUnited States (Senate and House of \nRepresentatives)\nData was collected using the advanced search feature \nof the U.S. Congressional Record website. MGI searched \nthe terms “Artificial Intelligence” and “Machine Learning” \n(quotes included) and downloaded the results as a \nCSV. The “word variant” option was not selected, and \nproceedings included Senate, House of Representatives, \nand Extensions of Remarks, but did not include the Daily \nDigest. Data is as of Dec. 31, 2020, and data is available \nonline from the 104th Congress onward (1995).\nEach count indicates that Artificial Intelligence or Machine \nLearning was mentioned during a particular event \ncontained in the Congressional Record, including the \nreading of a bill. If a speaker mentioned AI or ML multiple \ntimes within remarks, or multiple speakers mentioned AI or \nML within the same event, it would appear only once as a \nresult. Counts for Artificial Intelligence or Machine Learning \nare separate, as they were conducted in separate searches. \nMentions of the abbreviations AI or ML are not included.\nU.S. AI POLICY PAPER\nSource\nData collection and analysis was performed by Stanford \nInstitute of Human-Centered Artificial Intelligence and AI Index.\nOrganizations\nTo develop a more nuanced understanding of the \nthought leadership that motivates AI policy, we tracked \npolicy papers published by 36 organizations across three \nbroad categories including: \nThink Tanks, Policy Institutes & Academia: This includes \norganizations where experts (often from academia and \nthe political sphere) provide information and advice \non specific policy problems. We included the following \n27 organizations: AI PULSE at UCLA Law, American \nEnterprise Institute, Aspen Institute, Atlantic Council, \nBerkeley Center for Long-Term Cybersecurity, Brookings \n\n\n29\nArtificial Intelligence\nIndex Report 2021\nCHAPTER 7 PREVIEW\nCHAPTER 7:  \nAI POLICY AND  \nNATIONAL STRATEGIES\nAPPENDIX\nInstitution, Carnegie Endowment for International Peace, \nCato Institute, Center for a New American Security, \nCenter for Strategic and International Studies, Council \non Foreign Relations, Georgetown Center for Security \nand Emerging Technology (CSET), Harvard Belfer Center, \nHarvard Berkman Klein Center, Heritage Foundation, \nHudson Institute, MacroPolo, MIT Internet Policy Research \nInitiative, New America Foundation, NYU AI Now Institute, \nPrinceton School of Public and International Affairs, RAND \nCorporation, Rockefeller Foundation, Stanford Institute \nfor Human-Centered Artificial Intelligence (HAI), Stimson \nCenter, Urban Institute, Wilson Center.\nCivil Society, Associations & Consortiums: Not-for profit \ninstitutions including community-based organizations \nand NGOs advocating for a range of societal issues. We \nincluded the following nine organizations: Algorithmic \nJustice League, Alliance for Artificial Intelligence in \nHealthcare, Amnesty International, EFF, Future of Privacy \nForum, Human Rights Watch, IJIS, Institute for Electrical \nand Electronics Engineers, Partnership on AI\nIndustry & Consultancy: Professional practices providing \nexpert advice to clients and large industry players. We \nincluded six prominent organizations in this space: Accenture, \nBain & Co., BCG, Deloitte, Google AI, McKinsey & Company\nMethodology\nEach broad topic area is based on a collection of underlying \nkeywords that describes the content of the specific paper. \nWe included 17 topics that represented the majority of \ndiscourse related to AI between 2019-2020. These topic \nareas and the associated keywords are listed below.\n• \u0007\nHealth & Biological Sciences: medicine, healthcare \nsystems, drug discovery, care, biomedical research, \ninsurance, health behaviors, COVID-19, global health\n• \u0007\nPhysical Sciences: chemistry, physics, astronomy, earth \nscience\n• \u0007\nEnergy & Environment: Energy costs, climate change, \nenergy markets, pollution, conservation, oil & gas, \nalternative energy\n• \u0007\nInternational Affairs & International Security: \ninternational relations, international trade, developing \ncountries, humanitarian assistance, warfare, regional \nsecurity, national security, autonomous weapons\n• \u0007\nJustice & Law Enforcement: civil justice, criminal justice, \nsocial justice, police, public safety, courts\n• \u0007\nCommunications & Media: social media, disinformation, \nmedia markets, deepfakes\n• \u0007\nGovernment & Public Administration: federal \ngovernment, state government, local government, public \nsector efficiency, public sector effectiveness, government \nservices, government benefits, government programs, \npublic works, public transportation\n• \u0007\nDemocracy: elections, rights, freedoms, liberties, \npersonal freedoms\n• \u0007\nIndustry & Regulation: economy, antitrust, M&A, \ncompetition, finance, management, supply chain, \ntelecom, economic regulation, technical standards, \nautonomous vehicle industry & regulation\n• \u0007\nInnovation & Technology: advancements and \nimprovements in AI technology, R&D, intellectual \nproperty, patents, entrepreneurship, innovation \necosystems, startups, computer science, engineering\n• \u0007\nEducation & Skills: early childhood, K-12, higher \neducation, STEM, schools, classrooms, reskilling\n• \u0007\nWorkforce & Labor: labor supply and demand, talent, \nimmigration, migration, personnel economics, future of \nwork\n• \u0007\nSocial & Behavioral Sciences: sociology, linguistics, \nanthropology, ethnic studies, demography, geography, \npsychology, cognitive science\n• \u0007\nHumanities: arts, music, literature, language, \nperformance, theater, classics, history, philosophy, \nreligion, cultural studies\n• \u0007\nEquity & Inclusion: biases, discrimination, gender, \nrace, socioeconomic inequality, disabilities, vulnerable \npopulations\n• \u0007\nPrivacy, Safety & Security: anonymity, GDPR, \nconsumer protection, physical safety, human control, \ncybersecurity, encryption, hacking\n• \u0007\nEthics: transparency, accountability, human \nvalues, human rights, sustainability, explainability, \ninterpretability, decision-making norms\n\n\nArtificial Intelligence\nIndex Report 2022\nCHAPTER 5: \nAI Policy and  \nGovernance\n\n\n2\nArtificial Intelligence\nIndex Report 2022\nOverview\t\n  3\nChapter Highlights\t\n  4\n5.1 AI AND POLICYMAKING\t\n  5\nGlobal Legislation Records on AI\t\n  5\n\t\nBy Geographic Area\t\n  6\n\t\nFederal AI Legislation in the \n\t\nUnited States\t\n  7\n\t\nHighlight: A Closer Look \n\t\nat the Legislation\t\n  8\nState-Level AI Legislation \nin the United States\t\n  9\n\t\nBy State\t\n 10\n\t\nSponsorship by Political Party\t\n 11\nMentions of AI in Legislative Records\t\n 12\n\t\nAI Mentions in U.S. Congressional \n\t\nRecords\t\n 12\n\t\nAI Mentions in Global Legislative \n\t\nProceedings\t\n 13\n\t\nBy Geographic Area\t\n 14\nU.S. AI Policy Papers\t\n 15\n\t\nBy Topic\t\n 16\n5.2 U.S. PUBLIC INVESTMENT IN AI\t\n 17 \nFederal Budget for Nondefense AI R&D\t\n 17\nU.S. Department of Defense \nBudget Request\t\n 18\n\t\nHighlight: DOD Top Five \n\t\nHighest-Funded Programs\t\n 19\n\t\nDOD AI R&D Spending by Department\t  20\nU.S. Government AI-Related  \nContract Spending\t\n 21\n\t\nTotal Contract Spending\t\n 21\n\t\nContract Spending by Department \n\t\nand Agency\t\n 22\n\t\nHighlight: Largest Contract for Five\n\t\nTop-Spending Departments in 2021\t\n 24\nAPPENDIX\t\n 25 \nChapter Preview\nCHAPTER 5:\nACCESS THE PUBLIC DATA\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n3\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nOverview\nAs AI has become an increasingly ubiquitous topic in the last decade, \nintergovernmental, national, and regional organizations have worked to \ndevelop policies and strategies around AI governance. These actors are \ndriven by the understanding that it is imperative to find ways to address \nthe ethical and societal concerns surrounding AI, while maximizing \nits benefits. Active and informed governance of AI technologies has \nbecome a priority for many governments around the world.\nThis chapter examines the intersection of AI and governance, and takes \na closer look at how governments in different countries, regions, and \nU.S. states are working to manage AI technologies. It begins by looking \nat AI policymaking across the globe and within the United States, \nexploring which countries and political actors are most keen to advance \nAI legislation, and what kind of AI subtopics, from privacy to ethics, are \nthe focus of most legislative attention. Then the chapter takes a deep \ndive into one of the world’s top public sector investors in AI, the United \nStates, and studies how much its various government departments have \nspent on AI in the past five years.\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n4\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nCHAPTER 5: AI POLICY AND GOVERNANCE\nCHAPTER HIGHLIGHTS\n• \u0007\nAn AI Index analysis of legislative records on AI in 25 countries shows that the number of bills \ncontaining “artificial intelligence” that were passed into law grew from just 1 in 2016 to 18 in \n2021. Spain, the United Kingdom, and the United States passed the highest number of AI-related \nbills in 2021, with each adopting three.\n• \u0007\nThe federal legislative record in the United States shows a sharp increase in the total number of \nproposed bills that relate to AI from 2015 to 2021, while the number of bills passed remains low, \nwith only 2% ultimately becoming law.\n• \u0007\nState legislators in the United States passed 1 out of every 50 proposed bills that contain AI \nprovisions in 2021, while the number of such bills proposed grew from 2 in 2012 to 131 in 2021.\n• \u0007\nIn the United States, the current congressional session (the 117th) is on track to record the greatest \nnumber of AI-related mentions since 2001, with 295 mentions by the end of 2021, half way \nthrough the session, compared to 506 in the previous (116th) session.\n\n\n5\nChapter 5 Preview\nGLOBAL LEGISLATION  \nRECORDS ON AI \n​\nGovernments and legislative bodies across the globe are \nincreasingly seeking to pass laws to provide funding for \nAI development and innovation, while also promoting the \nintegration of human-centered values. The AI Index has \nconducted an analysis of laws passed in 25 countries by \ntheir legislative bodies that contain the words “artificial \nintelligence” from 2016 to 2021.\nTaken together, the 25 countries analyzed have passed a \ntotal of 55 AI-related bills. Figure 5.2.1 demonstrates that in \nthe past six years, there has been a sharp increase in terms \nof the total number of AI-related bills passed into law.1\n5.1 AI AND POLICYMAKING\n1 Note that the analysis only includes laws passed by national legislative bodies (e.g.  congress, parliament) with the keyword “artificial intelligence” in various languages in the title or body of the bill \ntext. See the appendix for the methodology. Countries included: Australia, Belgium, Brazil, Canada, China, Denmark, Finland, France, Germany, India, Ireland, Italy, Japan, the Netherlands, New Zealand, \nNorway, Russia, Singapore, South Africa, South Korea, Spain, Sweden, Switzerland, the United Kingdom, and the United States.\nArtificial Intelligence\nIndex Report 2022\nDiscussions around AI governance regulation have accelerated over the past decade, resulting in policy proposals across various \nlegislative bodies. This section first examines AI-related legislation that has either been proposed or passed into law across different \ncountries and regions, followed by a focused analysis of state-level legislation in the United States. It then takes a closer look at \ncongressional and parliamentary records on AI across the world and concludes with data on the number of policy papers published \nin the United States.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n2016\n2017\n2018\n2019\n2020\n2021\n0\n5\n10\n15\nNumber of AI-Related Bills\n18\nNUMBER of AI-RELATED BILLS PASSED into LAW in 25 SELECT COUNTRIES, 2016–21\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.1\n\n\n6\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nBy Geographic Area\nFigure 5.1.2a shows the number of laws containing \nmentions of AI that were enacted in 2021. Spain, the \nUnited Kingdom, and the United States led, each passing \nthree. Figure 5.1.2b shows the total number of legislation \npassed in the past six years. The United States dominated \nthe list with 13 bills, starting in 2017 with 3 new laws \npassed each subsequent year, followed by Russia, \nBelgium, Spain, and the United Kingdom.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n0\n1\n2\n3\n4\nNumber of AI-Related Bills\nSpain\nUnited Kingdom\nUnited States\nBelgium\nRussia\nFrance\nGermany\nItaly\nJapan\nSouth Korea\n3\n3\n3\n2\n2\n1\n1\n1\n1\n1\nNUMBER of AI-RELATED BILLS PASSED into LAW in SELECT COUNTRIES, 2021\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.2a\nThe United States \ndominated the list with 13 \nbills, starting in 2017 with \n3 new laws passed each \nsubsequent year, followed \nby Russia, Belgium, Spain, \nand the United Kingdom.\n\n\n7\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nFederal AI Legislation in the United States\nA closer look at the federal legislative record in the \nUnited States shows a sharp increase in the total number \nof proposed bills that relate to AI (Figure 5.1.3). In 2015, \njust one federal bill was proposed, while in 2021, there \nwere 130. Although this jump is significant, the number \nof bills related to AI being passed has not kept pace with \nthe growing volume of proposed AI-related bills. This gap \nwas most evident in 2021, when only 2% of all federal-\nlevel AI-related bills were ultimately passed into law. \n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n2015\n2016\n2017\n2018\n2019\n2020\n2021\n0\n20\n40\n60\n80\n100\n120\nNumber of AI-Related Bills\n130, Proposed\n3, Passed\nNUMBER of AI-RELATED BILLS in the UNITED STATES, 2015–21 (PROPOSED vs. PASSED)\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.3\n0\n1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\nNumber of AI-Related Bills\nUnited States\nRussia\nBelgium\nSpain\nUnited Kingdom\nFrance\nItaly\nSouth Korea\nJapan\nChina\nBrazil\nCanada\nGermany\nIndia\n13\n6\n4\n4\n4\n3\n5\n5\n5\n2\n1\n1\n1\n1\nNUMBER of AI-RELATED BILLS PASSED into LAW in SELECT COUNTRIES, 2016–21 (SUM)\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.2b\n\n\n8\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\nA Closer Look at the Legislation\nThe following subsection delves into some of the AI-related legislation passed into law since 2016. \nTable 5.1.1 demonstrates the wide range of AI-related issues that have piqued policymakers’ interest.\nCountry\nYear Passed\nBill Name\nDescription\nCanada\n2017\nBudget Implementation Act 2017, No. 1\nA provision of this act authorized the Canadian \ngovernment to make a payment of $125 million \nto the Canadian Institute for Advanced Research \nto support the development of a pan-Canadian \nartificial intelligence strategy. \nChina\n2019\nLaw of the People’s Republic of China \non Basic Medical and Health Care and \nthe Promotion of Health\nA provision of this law aimed to promote the \napplication and development of big data and \nartificial intelligence in the health and medical field \nwhile accelerating the construction of medical and \nhealthcare information infrastructure, developing \ntechnical standards on the collection, storage, \nanalysis, and application of medical and health data.\nRussia\n2020\nFederal Law of 24 April 2020 No. \n123-FZ on the Experiment to Establish \nSpecial Regulation in order to Create \nthe Necessary Conditions for the \nDevelopment and Implementation of \nArtificial Intelligence Technologies in \nthe Region of the Russian Federation \n– Federal City of Moscow and \nAmending the Articles 6 and 10 of the \nFederal Law on Personal Data\nThis law established an experimental framework \nfor the development and implementation of AI as \na five-year experiment to start in Moscow in July \n1, 2020, including allowing AI systems to process \nanonymized personal data for governmental and \ncertain commercial business activities.\nUnited Kingdom\n2020\nSupply and Appropriation (Main \nEstimates) Act 2020, c.13\nA provision of this act authorized the Office of \nQualifications and Examination Regulation to \nexplore opportunities for using artificial intelligence \nto improve the marking and administration of high-\nstakes qualifications.\nUnited States\n2020\nIOGAN ACT: Identifying Outputs of \nGenerative Adversarial Networks Act\nThis act directed the National Science Foundation \nto support research dedicated to studying the \noutputs of generative adversarial networks \n(deepfakes) and other comparable technologies.\nBelgium\n2021\nDecree on coaching and solution-\noriented support for job seekers, N. \n327\nA provision of this act directs the government \nto create an advisory group called the Ethics \nCommittee, which is responsible for submitting \nadvice if artificial intelligence tools are to be used \nfor digitization activities.\nFrance\n2021\nLaw N:2021-1485 of November \n15, 2021, aimed at reducing the \nenvironmental footprint of digital \ntechnology in France\nThis act sets up a monitoring system to evaluate \nenvironmental impacts of newly emerging digital \ntechnologies, in particular, artificial intelligence. \nTable 5.1.1\n\n\n9\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nSTATE-LEVEL AI LEGISLATION IN \nTHE UNITED STATES\nGrowing policy interest in AI can also be seen in the \nlarge number of AI-related bills recently proposed \nat the state level in the United States, based on data \nprovided by Bloomberg Government since 2012. \nBloomberg Government classified a bill as relating to \nAI if it contained AI-related keywords such as artificial \nintelligence, machine learning, or algorithmic bias.\nAs is the case on the federal level, there has been a \nsignificant increase in the number of AI bills proposed \nat the state level in the last decade (Figure 5.1.4). \nIn 2012, the first two pieces of AI-related legislation \nwere proposed when New Jersey assembly member \nAnnette Quijano directed the New Jersey Motor Vehicle \nCommission to establish driver’s license endorsements \nfor autonomous vehicles. In the past 10 years, the \nincrease has been substantial, from 2 bills in 2012 to 131 \nin 2021.\nA notable difference between AI-related lawmaking in the \nUnited States on the federal versus the state level is that \na greater proportion of proposed state-level AI bills have \nactually passed. In 2021, of the 131 proposed state bills, \n26 were passed into law (20%), or 1 out of 5 proposed \nbills became law. This ratio is significantly higher when \ncompared to the federal level, where 1 out of every 50 \nproposed bills became law in 2021.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n2021\n0\n20\n40\n60\n80\n100\n120\n140\nNumber of AI-Related Bills\n103\n66\n26\n25\n74\n10\n13\n12\n17\n9\n8\n2\n10\n14\n9\n9\n29\n26\n87\n77\n131\nNUMBER of STATE-LEVEL AI-RELATED BILLS in the UNITED STATES, 2012–21\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.4\nPassed\nProposed\nVetoed\nA notable difference between \nAI-related lawmaking in the \nUnited States on the federal \nversus the state level is \nthat a greater proportion of \nproposed state-level AI bills \nhave actually passed.\n\n\n10\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nBy State\nIn the United States, AI \nlawmaking has been relatively \nwidespread across all states. As \nof 2021, 41 out of 50 states have \nproposed at least one AI-related \nbill, but certain states have been \nparticularly active in generating \nAI legislation. Figure 5.1.5 shows \nthat Massachusetts has proposed \nthe most AI bills, with 40 since \n2012, followed by Hawaii (35) \nand New Jersey (32). Focusing \non just 2021 in Figure 5.1.6, \nMassachusetts was the state that \nproposed the most AI-related \nbills, with 20, followed by Illinois \n(15) and Alabama (12).\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\nMO\n4\nNM\n1\nMN\n2\nWA\n14\nMD\n8\nWV\n9\nMA\n40\nWY\n0\nCO\n2\nOH\n3\nMS\n7\nMT\n0\nME\n0\nNC\n6\nNH\n0\nND\n0\nOK\n2\nDC\n8\nGA\n3\nCA\n29\nOR\n1\nNV\n10\nNY\n31\nAK\n0\nTN\n7\nVA\n8\nNE\n2\nSC\n1\nCT\n2\nAZ\n7\nAR\n1\nSD\n0\nDE\n1\nNJ\n32\nPA\n7\nKY\n1\nWI\n0\nUT\n3\nKS\n1\nVT\n8\nLA\n0\nAL\n21\nMI\n3\nTX\n17\nHI\n35\nFL\n22\nIL\n28\nIN\n1\nID\n0\nIA\n1\nRI\n5\nNUMBER of STATE-LEVEL PROPOSED AI-RELATED BILLS in the UNITED\nSTATES by STATE, 2012–21 (SUM)\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.5\nMO\n1\nNM\n0\nMN\n0\nWA\n6\nMD\n4\nWV\n3\nMA\n20\nWY\n0\nCO\n2\nOH\n1\nMS\n3\nMT\n0\nME\n0\nNC\n3\nNH\n0\nND\n0\nOK\n1\nDC\n6\nGA\n0\nCA\n4\nOR\n1\nNV\n1\nNY\n8\nAK\n0\nTN\n2\nVA\n1\nNE\n0\nSC\n1\nCT\n0\nAZ\n0\nAR\n0\nSD\n0\nDE\n0\nNJ\n4\nPA\n3\nKY\n0\nWI\n0\nKS\n0\nUT\n2\nVT\n3\nLA\n0\nAL\n12\nMI\n0\nTX\n6\nFL\n7\nIN\n0\nHI\n7\nID\n0\nIA\n1\nIL\n15\nRI\n3\nNUMBER of STATE-LEVEL PROPOSED AI-RELATED BILLS in the UNITED\nSTATES by STATE, 2021\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.6\n\n\n11\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nSponsorship by Political Party\nState-level AI legislation data reveals that there is a \npartisan dynamic to AI lawmaking. Figure 5.1.7 plots the \nnumber of AI-related bills sponsored at the state level by \nDemocratic and Republican lawmakers. Although there \nhas been an increase in AI bills proposed by members \nof both parties since 2012, in the past four years, the \ndata suggests Democrats were more likely to sponsor \nAI-related legislation. Whereas Democrats sponsored \nonly two more AI bills than Republicans in 2018, they \nsponsored 39 more in 2021.\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n2021\n0\n10\n20\n30\n40\n50\n60\n70\n80\nNumber of AI-Related Bills\n79, Democratic\n40, Republican\nNUMBER of STATE-LEVEL PROPOSED AI-RELATED BILLS in the UNITED STATES by SPONSOR PARTY, 2012–21\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.7\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n12\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nMENTIONS OF AI IN LEGISLATIVE \nRECORDS\nAnother barometer of legislative interest in AI is the \nnumber of mentions of “artificial intelligence” in \ngovernmental and parliamentary proceedings. This \nsubsection considers data on mentions of AI both \nin U.S. congressional records and the parliamentary \nproceedings of other countries based on AI Index and \nBloomberg Government data. \nAI Mentions in U.S. Congressional Records\nIn the last five years, and especially in 2021, U.S. \ncongressional sessions have devoted increasing amounts \nof time to discussions of AI. This section presents data \nfrom Bloomberg Government concerning mentions of AI-\nrelated keywords in congressional proceedings, broken \ndown by legislation, congressional committee reports, \nand congressional research service reports.\nAccording to Figure 5.1.8, the current congressional \nsession (the 117th) is on track (as of the end of 2021) \nto record the greatest number of AI-related mentions \nsince 2001. The most recently completed congressional \nsession, the 116th (2019-2020), saw 506 AI mentions, \nnearly 3.4 times as many mentions as there were during \nthe 115th session (2017–2018), and 30 times as many as \nthe 114th session (2015–2016).\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n107th\n(2001–02)\n108th\n(2003–04)\n109th\n(2005–06)\n110th\n(2007–08)\n111th\n(2009–10)\n112th\n(2011–12)\n113th\n(2013–14)\n114th\n(2015–16)\n115th\n(2017–18)\n116th\n(2019–20)\n117th\n(2021–)\n0\n100\n200\n300\n400\n500\nNumber of Mentions\n245\n139\n129\n178\n66\n44\n39\n83\n27\n4\n7\n25\n17\n18\n17\n12\n17\n149\n506\n295\nMENTIONS of AI in the U.S. CONGRESSIONAL RECORD by LEGISLATIVE SESSION, 2001–21\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.8\nLegislation\nCongressional Research Service Reports\nCommittee Reports\n\n\n13\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nAI Mentions in Global Legislative Proceedings\nAI mentions in governmental proceedings are on the \nrise not only in the United States but also in many other \ncountries across the world. The AI Index conducted an \nanalysis on the minutes or proceedings of legislative \nsessions in 25 countries that contain the keyword \n“artificial intelligence” from 2016 to 2021. Figure 5.1.9 \nshows that the mentions of AI in legislative proceedings \nin 25 select countries grew 7.7 times in the past six years.2\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n2016\n2017\n2018\n2019\n2020\n2021\n0\n200\n400\n600\n800\n1,000\n1,200\nNumber of Mentions\n1,323\nNUMBER of MENTIONS of AI in LEGISLATIVE PROCEEDINGS in 25 SELECT COUNTRIES, 2016–21\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.9\n2 See the appendix for the methodology. Countries included: Australia, Belgium, Brazil, Canada, China, Denmark, Finland, France, Germany, India, Ireland, Italy, Japan, the Netherlands, New Zealand, \nNorway, Russia, Singapore, South Africa, South Korea, Spain, Sweden, Switzerland, the United Kingdom, and the United States.\n\n\n14\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nBy Geographic Area\nFigure 5.1.10a shows the number of legislative \nproceedings containing mentions of AI that were \nenacted in 2021. Similar to the trend in the number of \nAI mentions in bills passed into laws, Spain, the United \nKingdom, and the United States topped the list. Figure \n5.1.2b shows the total number of AI mentions in the past \nsix years. The United Kingdom dominated the list with \n939 mentions, followed by Spain, Japan, the United \nStates, and Australia.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n0\n50\n100\n150\n200\n250\n300\nNumber of Mentions\nSpain\nUnited Kingdom\nUnited States\nAustralia\nJapan\nIreland\nBrazil\nItaly\nSingapore\nBelgium\nGermany\nFrance\nCanada\nNorway\nSweden\nFinland\nRussia\nSouth Africa\nNetherlands\nIndia\nNew Zealand\nSouth Korea\nDenmark\nSwitzerland\n269\n185\n132\n122\n60\n46\n64\n20\n95\n25\n76\n47\n22\n72\n10\n16\n12\n12\n11\n6\n6\n3\n5\n7\nNUMBER of MENTIONS of AI in LEGISLATIVE PROCEEDINGS in\nSELECT COUNTRIES, 2021\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.10a\n0\n200\n400\n600\n800\n1000\nNumber of Mentions\nUnited Kingdom\nSpain\nJapan\nUnited States\nAustralia\nSingapore\nIreland\nItaly\nGermany\nFrance\nBrazil\nBelgium\nCanada\nFinland\nSweden\nNetherlands\nRussia\nNorway\nIndia\nSouth Africa\nDenmark\nNew Zealand\nSouth Korea\nSwitzerland\n466\n939\n559\n422\n282\n222\n410\n164\n120\n158\n155\n123\n34\n58\n33\n52\n67\n111\n78\n27\n27\n15\n21\n71\nNUMBER of MENTIONS of AI in LEGISLATIVE PROCEEDINGS in\nSELECT COUNTRIES, 2016–2021 (SUM)\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.10b\n\n\n15\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nU.S. AI POLICY PAPERS\nTo estimate activities outside national governments \nthat are also informing AI-related rulemaking, the \nAI Index tracks 55 U.S.-based organizations that \npublished policy papers in the past four years. Those \norganizations include: think tanks and policy institutes \n(19); university institutes and research programs (14); \ncivil society organizations, associations, and consortiums \n(9); industry and consultancy organizations (9); and \ngovernment agencies (4).3 A policy paper in this section \nis defined as a research paper, research report, brief, or \nblog post that addresses issues related to AI and makes \nspecific recommendations to policymakers. Topics of \nthose papers are divided into primary and secondary \ncategories: A primary topic is the main focus of the paper, \nwhile a secondary topic is a subtopic of the paper or an \nissue that was briefly explored.\nFigure 5.1.11 plots the total number of U.S.-based AI-\nrelated policy papers that have been published from \n2018 to 2021, which can proxy the general interest in AI \nwithin the U.S. policymaking space. The total number of \npolicy papers has tripled since 2018, peaking in 2020 with \n273, and decreasing slightly in 2021, with 210.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\n2018\n2019\n2020\n2021\n0\n50\n100\n150\n200\n250\nNumber of Policy Papers\n210\nNUMBER of AI-RELATED POLICY PAPERS by U.S.-BASED ORGANIZATIONS, 2018–21\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nFigure 5.1.11\n3 The complete list of organizations the Index followed can be found in the Appendix.\n\n\n16\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nBy Topic\nIn 2021, the leading primary topics were Privacy, Safety, \nand Security; Innovation and Technology; and Ethics \n(Figure 5.1.12). Certain topics, such as government and \npublic administration, education and skills, as well as \ndemocracy, did not feature prominently as primary \ntopics, but they were reported on more frequently \nas secondary topics. Among the AI topics to receive \ncomparatively little attention from tracked organizations \nare those that relate to energy and the environment, \nhumanities, physical sciences, and social and behavioral \nsciences.\n5.1 AI and Policymaking\nCHAPTER 5: AI POLICY AND GOVERNANCE\nPrimary Topic\nSecondary Topic\n0\n20\n40\n60\n0\n20\n40\n60\nPrivacy, Safety, and Security\nInnovation and Technology\nEthics\nInt'l A\"airs and Int'l Security\nIndustry and Regulation\nEquity and Inclusion\nWorkforce and Labor\nGov't and Public Administration\nJustice and Law Enforcement\nEducation and Skills\nCommunications and Media\nHealth and Biological Sciences\nSocial and Behavioral Sciences\nDemocracy\nPhysical Sciences\nEnergy and Environment\nHumanities\n36\n59\n34\n29\n62\n62\n33\n23\n51\n51\n15\n0\n4\n2\n7\n1\n1\n30\n63\n36\n45\n45\n29\n58\n58\n57\n13\n51\n17\n17\n3\n3\n1\n1\nNUMBER of AI-RELATED POLICY PAPERS by U.S.-BASED ORGANIZATIONS by TOPIC, 2021\nSource: AI Index, 2021 | Chart: 2022 AI Index Report\nNumber of Policy Papers\nFigure 5.1.12\n\n\n17\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nFEDERAL BUDGET FOR NONDEFENSE  \nAI R&D \nIn December 2021, the National Science and Technology Council \npublished a report on the public-sector AI R&D budget across \ndepartments and agencies participating in the Networking and \nInformation Technology Research and Development (NITRD) \nprogram and the National Artificial Intelligence Initiative. The report \ndoes not include information on classified AI R&D investment by the \ndefense and intelligence agencies.\nIn fiscal year (FY) 2021, nondefense U.S. government agencies \nallocated a total of $1.53 billion to AI R&D spending, approximately \n2.7 times what was spent in FY 2018 (Figure 5.2.1). This figure \nis projected to rise 8.8% for FY 2022, with a total of $1.67 billion \nrequested.4 The increasing amount spent on AI R&D by nondefense \ndepartments indicates the U.S. government’s continued strong \ninterest in public sector funding for AI research and development \nspanning a wide range of federal agencies.\n5.2 U.S. PUBLIC INVESTMENT IN AI \n4 See NITRD website for details on AI R&D investment FY 2018-22 with the breakdown of core AI vs AI crosscut. Note that AI crosscutting budget data is not available for FY 2018.\nArtificial Intelligence\nIndex Report 2022\nThis section examines the public AI investment in the United States, based on data from the U.S. government and Bloomberg Government.\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\nFY18 (ENACTED)\nFY19 (ENACTED)\nFY20 (ENACTED)\nFY21 (ENACTED)\nFY22 (REQUESTED)\n0.00\n0.50\n1.00\n1.50\nBudget (in billions of U.S. Dollars)\n0.56\n1.43\n1.53\n1.67\n1.11\nU.S. FEDERAL BUDGET for NONDEFENSE AI R&D, FY 2018–22\nSource: U.S. NITRD Program, 2022 | Chart: 2022 AI Index Report\nFigure 5.2.1\nThe increasing amount \nspent on AI R&D by \nnondefense departments \nindicates the U.S. \ngovernment’s continued \nstrong interest in \npublic sector funding \nfor AI research and \ndevelopment spanning \na wide range of federal \nagencies.\n\n\n18\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nU.S. DEPARTMENT OF DEFENSE \nBUDGET REQUEST\nSpending on AI by the U.S. Department of Defense (DOD) \ncan be proxied by looking at the publicly available \nrequests made by the DOD for research, development, \ntest, and evaluation (RDT&E) relating to AI. In FY 2021, \nDOD allocated $9.26 billion across 500 AI R&D programs \n(Figure 5.2.2), a 6.68% increase from the $8.68 billion \nspent in 2020. For FY 2022, the department has requested \n$10 billion so far, which is likely to grow once additional \nrequests and congressional appropriations are taken into \naccount.\nImportant data caveat: This chart is indicative of one \nof the challenges of quantifying public AI spending. \nBloomberg Government’s analysis that searches AI-\nrelevant keywords in DOD budgets shows that the \ndepartment is requesting $10.0 billion for AI-specific R&D \nin FY 2022. However, DOD’s own measurement produces \na smaller number of $874 million. The discrepancy \nmay result from the difference in defining AI-related \nbudget items. For example, a research project that uses \nAI for cyber defense may count human, hardware, and \noperations-related expenditures within the AI-related \nbudget request, though the AI software component will \nbe much smaller.\nSum of FY20 Funding\nSum of FY21 Funding\nSum of FY22 Funding\n0\n2\n4\n6\n8\n10\nBudget (in billions of U.S. Dollars)\n10.00\n8.68\n9.26\n0.84: DOD Reported Budget on AI R&D\n0.93: DOD Reported Budget on AI R&D\n0.87: DOD Reported Budget on AI R&D\nU.S. DOD BUDGET for AI-SPECIFIC RESEARCH, DEVELOPMENT, TEST and EVALUATION (RDT&E), FY 2020–22\nSource: Bloomberg Government and U.S. Department of Defense, 2021 | Chart: 2022 AI Index Report\nFigure 5.2.2\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n19\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nCHAPTER 5: AI POLICY AND GOVERNANCE\nDOD Top Five Highest-Funded Programs\nThis section highlight offers a more qualitative look at some of the AI-related research projects the \nDOD prioritizes. Table 5.2.1 presents the five DOD-related AI programs that received the greatest \nfunding in 2021. In the past year, the DOD was interested in deploying AI for a number of purposes, \nfrom geospatial monitoring to reducing the threat posed by weapons of mass destruction. \nProgram Name\nDepartment\nFunds Received \n(in millions)\nPurpose\n1 Rapid Capability \nDevelopment and Maturation\nArmy\n257 \nFund the development, engineering, acquisition, \nand operation of various AI-related technological \nprototypes that could be used for military purposes. \n2 Counter Weapons of \nMass Destruction Advanced \nTechnology Development\nDefense Threat \nReduction \nAgency\n254 \nDevelop technologies that could “deny, defeat and \ndisrupt” weapons of mass destruction (WMD).\n3 Algorithmic Warfare  \nCross-Functional Teams –\nSoftware Pilot Program \nOffice of the \nSecretary of \nDefense\n230 \nAccelerate the integration of AI technologies in DOD \nsystems to “improve warfighting speed and lethality.”\n4 Joint Artificial Intelligence \nCenter\nDefense \nInformation \nSystems \nAgency\n137 \nDevelop, test, prototype, and demonstrate various AI \nand machine learning capabilities with the intention \nof integrating these capabilities across numerous \ndomains which include “supply chain, personal \nrecovery, infrastructure assessment, geospatial \nmonitoring during disaster and cyber sense making.”\n5 High Performance \nComputing Modernization \nProgram\nArmy\n96 \nInvestigate, demonstrate, and mature both general \nand special-purpose supercomputing environments \nthat are used to satisfy wide-ranging DOD priorities.\nTable 5.2.1\n5.2 U.S. Public Investment in AI\n\n\n20\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nDOD AI R&D Spending by Department\nDOD spending on AI R&D can also be broken down on \na subdepartmental level, which reveals how individual \ndefense agencies—the Army and the Navy, for instance—\ncompare in their AI spending (Figure 5.2.3). The U.S. \nNavy was the top-spending DOD agency in FY 2021 and \nis poised to maintain that position in 2022. They have \nrequested a total of $1.86 billion in FY 2022 for AI-related \nprojects, followed by the Army ($1.77 billion), the Office \nof the Secretary of Defense ($1.1 billion) and the Air Force \n($883 million).\nFY20 (ENACTED)\nFY21 (ENACTED)\nFY22 (REQUESTED)\n0\n2\n4\n6\n8\n10\nBudget (in billions of U.S. Dollars)\n1.00\n1.64\n1.86\n1.93\n1.63\n1.54\n1.92\n1.52\n1.75\n1.57\n1.72\n1.77\n1.19\n1.16\n1.18\n1.13\n1.12\n1.71\nU.S. DOD BUDGET for AI-SPECIFIC RESEARCH, DEVELOPMENT, TEST and EVALUATION (RDT&E) by\nDEPARTMENT, FY 2020–22\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nAir Force\nArmy\nDARPA\nDISA\nNavy\nOSD\nOther\nFigure 5.2.3\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n21\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nU.S. GOVERNMENT AI-RELATED \nCONTRACT SPENDING\nPublic investment in AI can also be measured by federal \ngovernment spending on AI-related contracts. U.S. \ngovernment agencies often award contracts to private \ncompanies for the supply of various goods and services \nthat typically occupy the largest share of an agency’s \nbudget. Bloomberg Government built a model to classify \nwhether a U.S. government contract was AI-related by \nadding up all contracting transactions that contain a set \nof more than 100 AI-specific keywords in their titles or \ndescriptions.5\nTotal Contract Spending\nIn 2021, federal departments and agencies spent a total of \n$1.79 billion on AI-related contracts. Although this amount \nis nearly double what was spent on AI-related contracts in \n2018 (roughly $920 million), it represents a slight decrease \nfrom the amount spent on AI-related contracts in 2020, \nwhich peaked at $1.97 billion (Figure 5.2.4).\n2000\n2001\n2002\n2003\n2004\n2005\n2006\n2007\n2008\n2009\n2010\n2011\n2012\n2013\n2014\n2015\n2016\n2017\n2018\n2019\n2020\n2021\n0.0\n0.5\n1.0\n1.5\n2.0\nContract Spending (in billions of U.S. Dollars)\n1.79\nU.S. GOVERNMENT TOTAL CONTRACT SPENDING on AI, FY 2000–21\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.2.4\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\n5 Note that contractors may add a number of keywords into their applications during the procurement process, so some of the projects included may have a relatively small AI component relative to \nother parts of technology.\n\n\n22\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nContract Spending by Department and Agency\nFigures 5.2.5 and 5.2.6 report AI-related contract spending \nby the top 10 federal agencies in 2021 and from 2000 to \n2021, respectively. The DOD outspent the rest of the U.S. \ngovernment on both charts by a significant margin. In \n2021, it spent $1.14 billion on AI-related contracts, roughly \nfive times what was spent by the next highest department, \nthe Department of Health and Human Services ($234 \nmillion). \nAggregate spending on AI contracts in the last four years \ntells a similar story. Since 2018, the DOD has spent $5.20 \nbillion on AI contracts, approximately seven times the next \nhighest spender, NASA ($1.41 billion). In fact, since 2018, \nthe DOD has spent twice as much on AI-related contracts \nas all other government agencies combined. Following the \nDOD and NASA are the Department of Health and Human \nServices ($700 million), the Department of Homeland \nSecurity ($362 million), and Department of the Treasury \n($156 million). \n0\n200\n400\n600\n800\n1000\n1200\nContract Spending (in millions of U.S. Dollars)\nDepartment of Defense (DOD)\nDepartment of Health and Human Services (HHS)\nNational Aeronautics and Space Administration (NASA)\nDepartment of Homeland Security (DHS)\nDepartment of Commerce (DOC)\nDepartment of the Treasury (TREAS)\nDepartment of Veterans A\"airs (VA)\nDepartment of Transportation (DOT)\nSecurities and Exchange Commission (SEC)\nDepartment of Agriculture (USDA)\nDepartment of Energy (DOE)\nAgency for International Development (USAID)\nDepartment of Justice (DOJ)\nDepartment of State (DOS)\nNational Science Foundation (NSF)\n1,138\n234\n159\n49\n38\n25\n81\n12\n12\n12\n6\n4\n8\n3\n2\nTOP CONTRACT SPENDING on AI by U.S. GOVERNMENT DEPARTMENT and AGENCY, 2021\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.2.5\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n23\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\n0\n1\n2\n3\n4\n5\nContract Spending (in billions of U.S. Dollars)\nDepartment of Defense (DOD)\nNational Aeronautics and Space Administration (NASA)\nDepartment of Health and Human Services (HHS)\nDepartment of Homeland Security (DHS)\nDepartment of the Treasury (TREAS)\nDepartment of Veterans A!airs (VA)\nDepartment of Commerce (DOC)\nDepartment of Agriculture (USDA)\nDepartment of Energy (DOE)\nSecurities and Exchange Commission (SEC)\nGeneral Services Administration (GSA)\nDepartment of State (DOS)\nSocial Security Administration (SSA)\nDepartment of Transportation (DOT)\n0.06\n0.06\n0.06\n0.06\n0.05\n0.05\n0.45\n0.07\n0.70\n0.32\n5.20\n0.15\n0.15\n1.41\nTOP CONTRACT SPENDING on AI by U.S. GOVERNMENT DEPARTMENT and AGENCY, 2000–21 (SUM)\nSource: Bloomberg Government, 2021 | Chart: 2022 AI Index Report\nFigure 5.2.6\n5.2 U.S. Public Investment in AI\nCHAPTER 5: AI POLICY AND GOVERNANCE\n\n\n24\nChapter 5 Preview\nArtificial Intelligence\nIndex Report 2022\nCHAPTER 5: AI POLICY AND GOVERNANCE\nLargest Contract for Five Top-Spending  \nDepartments in 2021\nTo paint a better picture of how different U.S. government departments use AI, Table 5.2.2 shows the \nmost expensive AI-related contract that the five highest AI-related-spending departments signed in \n2021. Last year, the U.S. government invested in AI to build autonomous vehicle prototypes, develop an \nAI imaging system that could assist with burn classification, and create robots capable of higher-level \nlunar navigation.\nContract Name\nDepartment\nAmount \n(in millions)\nPurpose\nPrototype Services in the Objective \nAreas of Automotive Cybersecurity, \nVehicle Safety Technologies, Vehicle \nLight Weighting, Autonomous Vehicles \nand Intelligent Systems, Connected \nVehicles, and Advanced Energy Storage \nTechnologies\nDOD\n70 \nTo acquire prototypes in the domain of \nautomotive cybersecurity, vehicle safety \ntechnologies, and autonomous vehicles and \nintelligent systems.\nBiomedical Advanced Research and \nDevelopment Authority (BARDA)\nHHS\n20 \nTo develop optical imaging devices and \nmachine learning algorithms to assist \nin classifying and healing wounds and \nconventional burns. \nCommercial Lunar Payload Services\nNASA\n14\nTo develop lunar robots capable of navigating \nthe moon’s south pole to acquire lunar \nresources and engage in lunar-based scientific \nactivities. \nSBIR-Autonomous Surveillance  \nTowers-Delivery Order\nDHS\n37\nTo construct towers capable of autonomous \nsurveillance. \nSchedule 70: Information Technology \nDOC\n13\nTo develop a prototype using AI technology \nthat can improve patent search. \nTable 5.2.2\n5.2 U.S. Public Investment in AI\n\n\n25\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nChapter 5: AI Policy and Governance\nAPPENDIX\nBLOOMBERG GOVERNMENT\nPrepared by Amanda Allen\nBloomberg Government is a premium, subscription-\nbased service that provides comprehensive information \nand analytics for professionals who interact with—or \nare affected by—the government. Delivering news, \nanalytics, and data-driven decision tools, Bloomberg \nGovernment’s digital workspace gives an intelligent edge \nto government affairs and contracting professionals. For \nmore information or a demo, visit about.bgov.com.\n \nMethodology\nContract Spending: Bloomberg Government’s Contracts \nIntelligence Tool structures all contracts data from \nwww.fpds.gov. The CIT includes a model of government \nspending on artificial intelligence-related contracts that is \nbased on a combination of government-defined product \nservice codes and more than 100 AI-related keywords. \nFor the section “U.S. Government Contract Spending,” \nBloomberg Government analysts used contract spending \ndata from fiscal year 2000 through fiscal year 2021.\nDefense RDT&E Budget: Bloomberg Government \norganized all the RDT&E budget request line items \navailable from the Defense Department Comptroller. For \nthe section “U.S. Department of Defense (DOD) Budget,” \nBloomberg Government used a set of AI-specific keywords \nto identify 500 unique budget activities related to artificial \nintelligence and machine learning worth a combined $5.9 \nbillion in FY 2021.\nLegislative Documents: Bloomberg Government \nmaintains a repository of congressional documents, \nincluding bills, Congressional Budget Office assessments, \nand reports published by congressional committees, \nthe Congressional Research Service, and other offices. \nBloomberg Government also ingests state legislative \nbills. For the section “AI Policy and Governance,” \nBloomberg Government analysts identified all legislation, \ncongressional committee reports, and CRS reports that \nreferenced one or more AI-specific keywords.\nAPPENDIX\n\n\n26\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nGLOBAL LEGISLATION RECORDS ON AI\nFor AI-related bills passed into laws, the AI Index performed searches of the keyword “artificial intelligence,” in respective \nlanguages, on the websites of 25 countries’ congresses or parliaments, in full-text of bills. Note that only laws passed \nby state-level legislative bodies and signed into law (i.e., by presidents or received royal assent) from 2015 to 2021 are \nincluded. Future AI Index reports hope to include analysis on other types of legal documents, such as regulations and \nstandards, adopted by state- or supranational-level legislative bodies, government agencies, etc.\nDenmark\nWebsite: https://www.retsinformation.dk/ \nKeyword: kunstig intelligen\nFilter:\n\t\n• Document Type: Laws\nFinland\nWebsite: https://www.finlex.fi/\nKeyword: tekoäly \nNoting under the Current Legislation section\nFrance \nWebsite: https://www.legifrance.gouv.fr/ \nKeyword: intelligence artificielle\nFilter: \n\t\n• texte consolidé\n\t\n• Document Type: Law\nGermany\nWebsite: http://www.gesetze-im-internet.de/index.html \nKeyword: künstliche Intelligenz\nFilter: \n\t\n• \u0007\nAll federal codes, statutes, and ordinances that are \ncurrently in force\n\t\n• \u0007\nVolltextsuche (full text)\n\t\n• \u0007\nUnd-Verknüpfung der Wörter (entire word)\nIndia \nWebsite: https://www.indiacode.nic.in \nKeyword: artificial intelligence\nNote: The website used allows for a search of keywords \nin legalization title but not in the full text, as such it is not \nuseful for this particular research. Therefore, a Google \nsearch using the “site” function to search the site with the \nkeyword of “artificial intelligence” is conducted.\nAustralia\nWebsite: www.legislation.gov.au \nKeyword: artificial Intelligence\nFilters: \n\t\n• Legislation types: Acts\n\t\n• \u0007\nPortfolios: Department of House of Representatives, \nDepartment of Senate\nNote: Texts in explanatory memorandum are not counted.\nBelgium\nWebsite: http://www.ejustice.just.fgov.be/loi/loi.htm  \nKeyword: intelligence artificielle\nBrazil\nWebsite: https://www.camara.leg.br/legislacao  \nKeyword: inteligência artificial\nFilter:\n\t\n• Federal legislation\n\t\n• Type: Law\nCanada\nWebsite: https://www.parl.ca/legisinfo/ \nKeyword: artificial Intelligence\nNote: Results were investigated to determine how many of \nthe bills introduced were eventually passed (i.e., received \nroyal assent) and bill status was recorded.\nChina\nWebsite: https://flk.npc.gov.cn/  \nKeyword: 人工智能\nFilters: \n\t\n• \u0007\nLegislative body: Standing Committee of the \nNational People’s Congress\nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n27\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nIreland\nWebsite: www.irishstatutebook.ie\nKeyword: artificial intelligence\nItaly\nWebsite: https://www.normattiva.it/\nKeyword: intelligenza artificiale\nFilter: \n\t\n• \u0007\nDocument Type: law\nJapan\nWebsite: https://elaws.e-gov.go.jp/\nKeyword: 人工知能\nFilter:\n\t\n• \u0007\nFull text\n\t\n• \u0007\nLaw\nNetherlands\nWebsite: https://www.overheid.nl/ \nKeyword: kunstmatige intelligentie\nFilter:\n\t\n• \u0007\nDocument Type: Wetten\nNew Zealand\nWebsite: www.legislation.govt.nz\nKeyword: Artificial intelligence\nFilter: \n\t\n• \u0007\nDocument type: acts\n\t\n• \u0007\nStatus option: For the status option (example: acts in \nforce, current bills, etc.) \nNorway\nWebsite: https://lovdata.no/\nKeyword: kunstig intelligens\nRussia\nWebsite: http://graph.garant.ru:8080/SESSION/PILOT/\nmain.htm (Database “The Federal Laws” in the official \nwebsite of the Federation Council of the Federal Assembly \nof the Russian Federation.)\nKeyword: искусственный интеллект\nFilter:\n\t\n• \u0007\nWords in text\nSingapore\nWebsite: https://sso.agc.gov.sg/\nKeyword: artificial intelligence\nFilter: \n\t\n• \u0007\nDocument Type: Current acts and subsidiary \nlegislation\nSouth Africa\nWebsite: www.gov.za\nKeyword: artificial intelligence\nFilter: \n\t\n• \u0007\nDocument: acts\nNote: This search function seemingly does not search \nwithin the context of the full text and so no results were \nreturned. Therefore, a Google search using the “site” \nfunction to search the site with the keyword of “artificial \nintelligence” is conducted. \nSouth Korea\nWebsite: https://law.go.kr/eng/; https://elaw.klri.re.kr/\nKeyword: artificial Intelligence or 인공 지능\nFilter:\n\t\n• \u0007\nType: Act \nNote: Cannot search combined words, so individual \nanalysis is conducted. \nSpain \nWebsite: https://www.boe.es/ \nKeyword: inteligencia artificial\nFilter:\n\t\n• \u0007\nType: law\n\t\n• \u0007\nHead of state (for passed laws)\nSweden\nWebsite: https://www.riksdagen.se/\nKeyword: artificiell intelligens\nFilter: Swedish Code of Statutes\nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n28\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nSwitzerland\nWebsite: https://www.fedlex.admin.ch/ \nKeyword: intelligence artificielle\nFilter: \n\t\n• \u0007\nText category: federal constitution, federal acts, and \nfederal decrees, miscellaneous texts, orders, and \nother forms of legislation.\n\t\n• \u0007\nPublication period for legislation was limited to \n2015-2021.  \nUnited Kingdom\nWebsite: https://www.legislation.gov.uk/\nKeyword: artificial intelligence\nFilter:\n\t\n• \u0007\nLegislation Type: U.K. Public General Acts & U.K. \nStatutory Instruments\nUnited States\nWebsite: https://www.congress.gov/ \nKeyword: artificial intelligence\nFilter:\n\t\n• \u0007\nSource: Legislation\nStatus of legislation: Became law\nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n29\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nMENTIONS OF AI IN AI-RELATED LEGISLATION PROCEEDINGS\nFor mentions of AI in AI-related legislative proceedings around the world, the AI Index performed searches of the keyword \n“artificial intelligence,” in respective languages, on the websites of 25 countries’ congresses or parliaments, usually under \nsections named “minutes,” “hansard,” etc.\nDenmark\nWebsite: https://www.retsinformation.dk/ \nKeyword: kunstig intelligens\nFilter:\n\t\n• Minutes\nFinland\nWebsite: https://www.eduskunta.fi/ \nKeyword: tiedot\nFilter:\n\t\n• Parliamentary Affairs and Documents\n\t\n• Public document: Minutes\n\t\n• Actor: Plenary sessions\nFrance \nWebsite: https://www.assemblee-nationale.fr/ \nKeyword: intelligence artificielle\nFilter: \n\t\n• Reports of the debates in session\nNote: Such documents were only prepared starting in \n2017.\nGermany\nWebsite: https://dip.bundestag.de/ \nKeyword: künstliche Intelligenz\nFilter: \n\t\n• Speeches, requests to speak in the plenum\nIndia \nWebsite: http://loksabhaph.nic.in/ \nKeyword: artificial intelligence\nFilter:\n\t\n• Exact word/phrase\nIreland\nWebsite: https://www.oireachtas.ie/ \nKeyword: artificial intelligence\nFilter: Content of parliamentary debates\nAustralia\nWebsite: https://www.aph.gov.au/Parliamentary_Business/\nHansard \nKeyword: artificial intelligence\nBelgium\nWebsite: http://www.parlement.brussels/search_form_fr/ \nKeyword: intelligence artificielle\nFilter\n\t\n• Document Type: all\nBrazil\nWebsite: https://www2.camara.leg.br/atividade-legislativa/\ndiscursos-e-notas-taquigraficas  \nKeyword: inteligência artificial\nFilter:\n\t\n• Federal legislation\n\t\n• Type: Law\nCanada\nWebsite: https://www.ourcommons.ca/PublicationSearch/\nen/?PubType=37 \nKeyword: artificial Intelligence\nChina\nWebsite: Various reports on the work of the government\nKeyword: 人工智能\nNote: The National People’s Congress is held once per \nyear and does not provide full legislative proceedings. \nHence, the counts included in the analysis only searched \nthe mentions of artificial intelligence in the only public \ndocument released from the Congress meetings, the \nReport on the Work of the Government, delivered by the \nPremier. \nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n30\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nItaly\nWebsite: https://aic.camera.it/aic/search.html \nKeyword: intelligenza artificiale\nFilter: \n\t\n• Type: All\n\t\n• Search by exact phrase\nJapan\nWebsite: https://kokkai.ndl.go.jp/#/  \nKeyword: 人工知能\nFilter:\n\t\n• Full text\n\t\n• Law\nNetherlands\nWebsite: https://www.tweedekamer.nl/kamerstukken?pk_\ncampaign=breadcrumb  \nKeyword: kunstmatige intelligentie\nFilter:\n\t\n• Parliamentary papers - Plenary reports\nNew Zealand\nWebsite: https://www.parliament.nz/en/pb/hansard-\ndebates/  \nKeyword: artificial intelligence \nNorway\nWebsite: https://www.stortinget.no/no/Saker-og-\npublikasjoner/Publikasjoner/Referater/ \nKeyword: kunstig intelligens\nNote: This search function does not directly allow the \nkeyword within minutes. Therefore, a Google search using \nthe “site” function to search the site with the keyword of \n“artificial intelligence” is conducted. \nRussia\nWebsite: http://transcript.duma.gov.ru/ \nKeyword: искусственный интеллект\nFilter:\n\t\n• Words in text\nSingapore\nWebsite: https://sprs.parl.gov.sg/search/home \nKeyword: artificial intelligence\nSouth Africa\nWebsite: https://www.parliament.gov.za/hansard \nKeyword: artificial intelligence\nNote: This search function does not search within the \ncontext of the full text and so no results were returned. \nTherefore, a Google search using the “site” function \nto search https://www.parliament.gov.za/storage/\napp/media/Docs/hansard/ with the keyword “artificial \nintelligence” is conducted. \nSouth Korea\nWebsite: http://likms.assembly.go.kr/ \nKeyword: 인공 지능\nFilter:\n\t\n• Meeting Type: All\nSpain \nWebsite: https://www.congreso.es/ \nKeyword: inteligencia artificial\nFilter:\n\t\n• Official publications of parliamentary proceedings\nSwitzerland\nWebsite: https://www.parlament.ch/ \nKeyword: intelligence artificielle\nFilter: \n\t\n• Parliamentary proceedings\nSweden\nWebsite: https://www.riksdagen.se/sv/global/\nsok/?q=&doktyp=prot\nKeyword: artificiell intelligens\nFilter:\n\t\n• Minutes\nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n31\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nUnited Kingdom\nhttps://hansard.parliament.uk/ \nKeyword: artificial intelligence\nFilter\n\t\n• References\nUnited States\nWebsite: https://www.congress.gov/ \nKeyword: artificial intelligence\nFilter:\n\t\n• Source: Congressional record\n\t\n• \u0007\nCongressional record section: Senate, House of \nRepresentatives, and Extensions of Remarks\nU.S. AI POLICY PAPERS\nOrganizations\nTo develop a more nuanced understanding of the thought \nleadership that motivates AI policy, we tracked policy \npapers published by 55 organizations in the United States \nor with a strong presence in the United States (expanded \nfrom the list of 36 organizations last year) across four \nbroad categories: \n\t\n• \u0007\nCivil Society, Associations & Consortiums: \nAlgorithmic Justice League, Alliance for Artificial \nIntelligence in Healthcare, Amnesty International, \nEFF, Future of Privacy Forum, Human Rights Watch, \nIJIS Institute, Institute for Electrical and Electronics \nEngineers, Partnership on AI\n\t\n• \u0007\nConsultancy: Accenture, Bain & Company, Boston \nConsulting Group, Deloitte, McKinsey & Company\n\t\n• \u0007\nGovernment Agencies: Congressional Research \nService, Library of Congress, Defense Technical \nInformation Center, Government Accountability \nOffice, Pentagon Library\n\t\n• \u0007\nPrivate Sector Companies: Google AI, Microsoft AI, \nNvidia, OpenAI\n\t\n• \u0007\nThink Tanks & Policy Institutes: American Enterprise \nInstitute, Aspen Institute, Atlantic Council, Brookings \nInstitute, Carnegie Endowment for International \nPeace, Cato Institute, Center for a New American \nSecurity, Center for Strategic and International \nStudies, Council on Foreign Relations, Heritage \nFoundation, Hudson Institute, MacroPolo, National \nSecurity Institute, New America Foundation, RAND \nCorporation, Rockefeller Foundation, Stimson \nCenter, Urban Institute, Wilson Center\n\t\n• \u0007\nUniversity Institutes & Research Programs: AI and \nHumanity Cornell University; AI Now Institute, \nNew York University; AI Pulse, UCLA Law; Belfer \nCenter for Science and International Affairs, \nHarvard University; Berkman Klein Center, Harvard \nUniversity; Center for Information Technology \nPolicy, Princeton University; Center for Long-Term \nCybersecurity, UC Berkeley; Center for Security \nand Emerging Technology, Georgetown University; \nCITRUS Policy Lab, UC Berkeley; Hoover Institution; \nInstitute for Human-Centered Artificial Intelligence, \nStanford University; Internet Policy Research \nInitiative, Massachusetts Institute of Technology; \nMIT Lincoln Laboratory; Princeton School of Public \nand International Affairs\nMethodology\nEach broad topic area is based on a collection of \nunderlying keywords that describe the content of the \nspecific paper. We included 17 topics that represented the \nmajority of discourse related to AI between 2018-2021. \nThese topic areas and the associated keywords are listed \nbelow:\n\t\n• \u0007\nHealth & Biological Sciences: medicine, healthcare \nsystems, drug discovery, care, biomedical research, \ninsurance, health behaviors, COVID-19, global health\n\t\n• \u0007\nPhysical Sciences: chemistry, physics, astronomy, \nearth science\n\t\n• \u0007\nEnergy & Environment: energy costs, climate \nchange, energy markets, pollution, conservation, oil \nand gas, alternative energy\n\t\n• \u0007\nInternational Affairs & International Security: \ninternational relations, international trade, \ndeveloping countries, humanitarian assistance, \nwarfare, regional security, national security, \nautonomous weapons\n\t\n• \u0007\nJustice & Law Enforcement: civil justice, criminal \njustice, social justice, police, public safety, courts\nChapter 5: AI Policy and Governance\nAPPENDIX\n\n\n32\nArtificial Intelligence\nIndex Report 2022\nChapter 5 Preview\nChapter 5: AI Policy and Governance\nAPPENDIX\n\t\n• \u0007\nCommunications & Media: social media, \ndisinformation, media markets, deepfakes\n\t\n• \u0007\nGovernment & Public Administration: federal \ngovernment, state government, local government, \npublic sector efficiency, public sector effectiveness, \ngovernment services, government benefits, \ngovernment programs, public works, public \ntransportation\n\t\n• \u0007\nDemocracy: elections, rights, freedoms, liberties, \npersonal freedoms\n\t\n• \u0007\nIndustry & Regulation: economy, antitrust, M&A, \ncompetition, finance, management, supply chain, \ntelecom, economic regulation, technical standards, \nautonomous vehicle industry and regulation\n\t\n• \u0007\nInnovation & Technology: advancements and \nimprovements in AI technology, R&D, intellectual \nproperty, patents, entrepreneurship, innovation \necosystems, startups, computer science, engineering\n\t\n• \u0007\nEducation & Skills: early childhood, K-12, higher \neducation, STEM, schools, classrooms, reskilling\n\t\n• \u0007\nWorkforce & Labor: labor supply and demand, talent, \nimmigration, migration, personnel economics, future \nof work\n\t\n• \u0007\nSocial & Behavioral Sciences: sociology, linguistics, \nanthropology, ethnic studies, demography, \ngeography, psychology, cognitive science\n\t\n• \u0007\nHumanities: arts, music, literature, language, \nperformance, theater, classics, history, philosophy, \nreligion, cultural studies\n\t\n• \u0007\nEquity & Inclusion: biases, discrimination, gender, \nrace, socioeconomic inequality, disabilities, \nvulnerable populations\n\t\n• \u0007\nPrivacy, Safety & Security: anonymity, GDPR, \nconsumer protection, physical safety, human control, \ncybersecurity, encryption, hacking\n\t\n• \u0007\nEthics: transparency, accountability, human \nvalues, human rights, sustainability, explainability, \ninterpretability, decision-making norms\n\n\n1 \n \n \n \n \n \n \n \nOn July 20, China's State Council issued a seminal document, entitled A Next \nGeneration Artificial Intelligence Development Plan. This important aspirational document \nsets out a top-level design blueprint charting the country's approach to developing artificial \nintelligence (AI) technology and applications, setting broad goals up to 2030. \n  \nPlease find the full text of the document below. \n \nThe translators produced analysis on the new document and Chinese AI ambitions for New \nAmerica here. \n  \nThe document has been translated into English by a group of experienced Chinese \nlinguists with deep backgrounds on the subject matter and on China's S&T establishment \nand current AI capabilities.  They are: Rogier Creemers, Leiden Asia Centre; Graham \nWebster, Yale Law School Paul Tsai China Center; Paul Triolo, Eurasia Group; and Elsa Kania. \n \nThe group is grateful to New America Cybersecurity Initiative Fellow John Costello for \ncomments that helped to improve the translation. \n   \nAny errors in translation are the responsibility of the translators, and we welcome \ncomments, which can be directed to the collaborators at this \naddress: chinacomments@newamerica.org \n \n \n \n\n\n2 \nState Council Notice on the Issuance of the Next \nGeneration Artificial Intelligence Development Plan \nCompleted: July 8, 2017 \nReleased: July 20, 2017 \n \nA Next Generation Artificial  \nIntelligence Development Plan \n \nThe rapid development of artificial intelligence (AI) will profoundly change human society \nand life and change the world. To seize the major strategic opportunity for the development \nof AI, to build China’s first-mover advantage in the development of AI, to accelerate the \nconstruction of an innovative nation and global power in science and technology, in \naccordance with the requirements of the CCP Central Committee and the State Council, this \nplan has been formulated. \nI.  The Strategic Situation \n \nThe development of AI has entered a new stage. After sixty years of evolution, especially in \nmobile Internet, big data, supercomputing, sensor networks, brain science, and other new \ntheories and new technologies, under the joint impetus of powerful demands of economic \nand social development, AI’s development has accelerated, displaying deep learning, \ncross-domain integration, man-machine collaboration, the opening of swarm intelligence, \nautonomous control, and other new characteristics. Big data-driven cognitive learning, \ncross-media collaborative processing, and man-machine collaboration–strengthened \nintelligence, swarm integrated intelligence, and autonomous intelligent systems have \nbecome the focus of the development of AI. The results of brain science research inspired \nhuman-like intelligence that awaits action; the trends involving the chips, hardware, and \nplatform have become apparent; the development of AI has entered into a new stage. At \npresent, the development a new generation of AI and related disciplines, theoretical \nmodeling, technological innovation, hardware and software upgrades, etc., all advance, \nprovoking chain-style breakthroughs, promoting the acceleration of the elevation of \neconomic and social domains from digitization and networkization to intelligentization.  \n \nAI has become a new focus of international competition. AI is a strategic technology that \nwill lead in the future; the world’s major developed countries are taking the development of \nAI as a major strategy to enhance national competitiveness and protect national security; \nintensifying the introduction of plans and strategies for this core technology, top talent, \nstandards and regulations, etc.; and trying to seize the initiative in the new round of \ninternational science and technology competition. At present, China’s situation in national \nsecurity and international competition is more complex, and [China] must, looking at the \nworld, take the development of AI to the national strategic level with systemic layout, take \nthe initiative in planning, firmly seize the strategic initiative in the new stage of \ninternational competition in AI development, to create new competitive advantage, \nopening up the development of new space, and effectively protecting national security. \n \nAI has become a new engine of economic development. AI has become the core driving \nforce for a new round of industrial transformation, [which] will advance the release of the \n\n\n3 \nhuge energy stored from the previous scientific and technological revolution and industrial \ntransformation, and create a new powerful engine, reconstructing production, distribution, \nexchange, consumption, etc., links in economic activities; with new demands taking shape \nfrom the macro to the micro within each domain of intelligentization; with the birth of new \ntechnologies, new products, new industries, new formats, new models; triggering \nsignificant changes in economic structure, profound changes in human modes of \nproduction, lifestyle, and thinking; and a whole leap of achieving social productivity. \nChina’s economic development enters a new normal, deepening the supply side of \nstructural reform task is very arduous, [and China] must accelerate the rapid application of \nAI, cultivating and expanding AI industries to inject new kinetic energy into China’s \neconomic development. \n \nAI brings new opportunities for social construction. China is currently in the decisive stage \nof comprehensively constructing a moderately prosperous society. The challenges of \npopulation aging, environmental constraints, etc., remain serious. The widespread use of AI \nin education, medical care, pensions, environmental protection, urban operations, judicial \nservices, and other fields will greatly improve the level of precision in public services, \ncomprehensively enhancing the people’s quality of life. AI technologies can accurately \nsense, forecast, and provide early warning of major situations for infrastructure facilities \nand social security operations; grasp group cognition and psychological changes in a \ntimely manner; and take the initiative in decision-making and reactions—which will \nsignificantly elevate the capability and level of social governance, playing an irreplaceable \nrole in effectively maintaining social stability. \n \nThe uncertainties in the development of AI create new challenges. AI is a disruptive \ntechnology with widespread influence that may cause: transformation of employment \nstructures; impact on legal and social theories; violations of personal privacy; challenges in \ninternational relations and norms; and other problems. It will have far-reaching effects on \nthe management of government, economic security, and social stability, as well as global \ngovernance. While vigorously developing AI, we must attach great importance to the \npotential safety risks and challenges, strengthen the forward-looking prevention and \nguidance on restraint, minimize risk, and ensure the safe, reliable, and controllable \ndevelopment of AI. \n \nChina possesses a favorable foundation for the development of AI. The nation has: \ndeployed the National Key Research and Development Plan’s key special projects, such as \nintelligent manufacturing; issued and implemented the “Internet +” and AI Three-Year \nActivities and Implementation Program, releasing a series of measures from science and \ntechnology research and development; and promoted applications and industrial \ndevelopment, and other aspects. As a result of many years of continuous accumulation, \nChina has achieved important progress in the field of AI, with the number of international \nscientific and technology papers published and the number of inventions patented ranked \nsecond in the world, while achieving important breakthroughs in certain domains of core \ncrucial technologies. Leading the world in voice recognition and visual recognition \ntechnologies; initially possessing the capability for leapfrog development in adaptive \nautonomous learning, intuitive sensing, comprehensive reasoning, hybrid intelligence, and \nswarm intelligence, etc.; with Chinese information processing, intelligent monitoring, \nbiometric identification, industrial robots, service robots, and unmanned driving gradually \nentering practical application; AI innovation and entrepreneurship have become \nincreasingly active, and a number of leading enterprises have accelerated their growth, \n\n\n4 \nreceiving widespread concern and recognition internationally. Accelerate the accumulation \nof technological capabilities and massive data resources, the organization integration of \nboth the huge demand for applications and an open market environment, which together \nconstitute China’s unique advantage in AI development. \n \nAt the same time, we must also clearly see that there is still a gap between China’s overall \nlevel of development of AI relative to that of developed countries—lacking major original \nresults in the basic theory, core algorithms, key equipment, high-end chips, major products \nand systems, foundational materials, components, software and interfaces, etc. Scientific \nresearch institutions and enterprises do not yet possess international influence upon \necological cycles and supply chain, lacking a systematic research and development layout; \ncutting-edge talent for AI is far from meeting demand. Adapting to the development of AI \nrequires the urgent improvement of basic infrastructure, policies and regulations, and \nstandards systems. \n \nFacing a new situation and new demands, we must take the initiative to pursue and adapt \nto change, firmly seize the major historic opportunity for the development of AI, stick \nclosely to development, study and evaluate the general trends, take the initiative to plan, \ngrasp the direction, seize the opportunity, lead the world in new trends in the development \nof AI, serve economic and social development, and support national security, promoting \nthe overall elevation of the nation’s competitiveness and leapfrog development. \n \nII.  The Overall Requirements \n(1)  Guiding Ideology \n  \nComprehensively implement the spirit of the 18th Party Congress and 18th Central \nCommittee’s Third, Fourth, Fifth, and Sixth Plenary Sessions. Thoroughly study and \nimplement the spirit of General Secretary Xi Jinping’s series of important sayings and new \nconcepts, new ideas, and new strategy for governing the country; according to the “five in \none” overall layout and “four comprehensives” strategic layout, conscientiously implement \nthe CPC Central Committee and State Council decision-making arrangements, deeply \nimplement the innovation-driven development strategy to accelerate the deep integration \nof AI with the economy, society and national defense as a primary line, to enhance: \nscientific and technological innovation capacity for a new generation of AI as the main \ndirection of attack; intelligent economy development; smart society construction; \nprotecting national security; building of knowledge clusters, technology clusters, and \nindustry clusters mutually integrated with talent, system, and culture, for a mutually \nsupporting ecosystem, advancing intelligentization as the center of humanity’s sustainable \ndevelopment. Comprehensively enhance society’s productive forces, comprehensive \nnational power, and national competitiveness, in order to provide strong support to \naccelerate the construction of an innovative new-type nation and global science and \ntechnology power, to achieve the two centennial goals and the great rejuvenation of the \nChinese nation. \n\n\n5 \n \n(2)  The Basic Principles \n  \nTechnology-Led. Grasp the global development trend of AI, highlight the deployment of \nforward-looking research and development, explore the layout in key frontier domains, \nlong-term support, and strive to achieve transformational and disruptive breakthroughs in \ntheory, methods, tools, and systems; comprehensively enhance original innovation \ncapability in AI, accelerate the construction of a first-mover advantage, to achieve high-\nend leading development. \n  \nSystems Layout. According to the different characteristics of foundational research, \ntechnological research and development, industrial development, and commercial \napplications, formulate a targeted systems development strategy. Fully give play to the \nadvantages of the socialist system to concentrate forces to do major undertakings, \npromote the planning and layout of projects, bases, and a talent pool, organically link \nalready-deployed major projects and new missions, continue current urgent needs and \nlong-term development echelons, construct innovation capacity, create a collaborative \nforce for institutional reforms and the policy environment. \n \nMarket-Dominant. Follow the rules of the market, remain oriented toward application, \nhighlight companies’ choices on the technological line and primary role in the development \nof commercial product standards, accelerate the commercialization of AI technology and \nresults, and create a competitive advantage. Grasp well the division of labor between \ngovernment and the market, better take advantage of the government in planning and \nguidance, policy support, security and guarding, market regulation, environmental \nconstruction, the formulation of ethical regulations, etc.  \n \nOpen-Source and Open. Advocate the concept of open-source sharing, and promote the \nconcept of industry, academia, research, and production units each innovating and in \nprincipal pursuing joint innovation and sharing. Follow the coordinated development law for \neconomic and national defense construction; promote two-way conversion and application \nfor military and civilian scientific and technological achievements and co-construction and \nsharing of military and civilian innovation resources; form an all-element, multi-domain, \nhighly efficient new pattern of civil-military integration. Actively participate in global \nresearch and development and management of AI, and optimize the allocation of \ninnovative resources on a global scale. \n(3)  Strategic Objectives \n  \nThese are divided into the following three steps: \n \nFirst, by 2020, the overall technology and application of AI will be in step with globally \nadvanced levels, the AI industry will have become a new important economic growth point, \nand AI technology applications will have become a new way to improve people’s \nlivelihoods, strongly supporting [China’s] entrance into the ranks of innovative nations and \ncomprehensively achieving the struggle toward the goal of a moderately prosperous \nsociety. \n\n\n6 \n  \n● By 2020 China will have achieved important progress in a new generation of AI \ntheories and technologies. It will have actualized important progress in big data \nintelligence, cross-medium intelligence, swarm intelligence, hybrid enhanced \nintelligence, and autonomous intelligence systems, and will have achieved \nimportant progress in other foundational theories and core technologies; the \ncountry will have achieved iconic advances in AI models and methods, core devices, \nhigh-end equipment, and foundational software. \n● The AI industry’s competitiveness will have entered the first echelon internationally. \nChina will have established initial AI technology standards, service systems, and \nindustrial ecological system chains. It will have cultivated a number of the world's \nleading AI backbone enterprises, with the scale of AI’s core industry exceeding 150 \nbillion RMB, and exceeding 1 trillion RMB as driven by the scale of related industries. \n● The AI development environment will be further optimized, opening up new \napplications in important domains, gathering a number of high-level personnel and \ninnovation teams, and initially establishing AI ethical norms, policies, and \nregulations in some areas. \n \nSecond, by 2025, China will achieve major breakthroughs in basic theories for AI, such that \nsome technologies and applications achieve a world-leading level and AI becomes the \nmain driving force for China’s industrial upgrading and economic transformation, while \nintelligent social construction has made positive progress. \n  \n● By 2025, a new generation of AI theory and technology system will be initially \nestablished, as AI with autonomous learning ability achieves breakthroughs in many \nareas to obtain leading research results.  \n● The AI industry will enter into the global high-end value chain. This new-generation \nAI will be widely used in intelligent manufacturing, intelligent medicine, intelligent \ncity, intelligent agriculture, national defense construction, and other fields, while \nthe scale of AI’s core industry will be more than 400 billion RMB, and the scale of \nrelated industries will exceed 5 trillion RMB.  \n● By 2025 China will have seen the initial establishment of AI laws and regulations, \nethical norms and policy systems, and the formation of AI security assessment and \ncontrol capabilities. \n \nThird, by 2030, China’s AI theories, technologies, and applications should achieve world-\nleading levels, making China the world’s primary AI innovation center, achieving visible \nresults in intelligent economy and intelligent society applications, and laying an important \nfoundation for becoming a leading innovation-style nation and an economic power.   \n \n● China will have formed a more mature new-generation AI theory and technology \nsystem. The country will achieve major breakthroughs in brain-inspired intelligence, \nautonomous intelligence, hybrid intelligence, swarm intelligence, and other areas, \nhaving important impact in the domain of international AI research and occupying \nthe commanding heights of AI technology. \n● AI industry competitiveness will reach the world-leading level. AI should be \nexpansively deepened and greatly expanded into production and livelihood, social \ngovernance, national defense construction, and in all aspects of applications, will \nbecome an expansive core technology for key systems, support platforms, and the \nintelligent application of a complete industrial chain and high-end industrial \n\n\n7 \nclusters, with AI core industry scale exceeding 1 trillion RMB, and with the scale of \nrelated industries exceeding 10 trillion RMB. \n● China will have established a number of world-leading AI technology innovation and \npersonnel training centers (or bases), and will have constructed more \ncomprehensive AI laws and regulations, and an ethical norms and policy system. \n(4)  Overall Deployment \n  \nThe development of AI is a complex systemic project related to the overall situation, that \nmust be arranged in accordance with “build one system, grasp the two attributes, adhere \nto the trinity, and strengthen the four supports” to form a strategic path for the healthy and \nsustainable development of AI. \n \nConstruct an open and cooperative AI technology innovation system. Target the weak \nfoundation in original theories, and the key difficulties and deficiencies in major products \nand systems. Establish foundational theories and a common technology system for a new \ngeneration of AI, laying out the construction of a major scientific and technological \ninnovation base. Strengthen the high-end talent team in AI to promote innovation and \ncooperative interactions. Form a continuous innovation capability for AI. \n \nGrasp AI’s characteristic high degree of integration of technological attributes and social \nattributes. It is necessary not only to increase efforts in the research and development and \napplications of AI, maximizing the potential of AI, but also to predict AI’s challenges, \ncoordinate industrial policies, innovate in policies and social policies, achieve the \ncoordination of encouraging development and reasonable regulation, and maximize risk \nprevention.  \n \nAdhere to the promotion of the trinity of breakthroughs in AI research and development, \nproduct applications, and fostering industry development. Adapt to the characteristics and \ntrends of AI development. Strengthen the deep integration of the innovation chain and \nindustrial chain, the interactive evolution of technology supply and market demand. Take \ntechnological breakthroughs to promote domain applications and industrial upgrading. \nThrough application demonstrations, promote the optimization of technologies and \nsystems. At the same time as greatly promoting technology applications and industrial \ndevelopment, strengthen long-term R&D layout and research. Achieve rolling development \nand continuous improvement. Ensure that theory is in the front, the technological \ncommanding heights are occupied, and applications are secure and controllable. \n  \nFully support science and technology, the economy, social development, and national \nsecurity.  Drive comprehensive elevation on national innovative capability with AI \ntechnological breakthroughs. Lead in the process of constructing a global science and \ntechnology power. Through strengthening intelligent industry and cultivating the intelligent \neconomy, create a new growth cycle for China’s next decade or even decades of economic \nprosperity. Through building an intelligent society, promote the improvement of people’s \nlivelihoods and welfare and implement people-centric development thinking. Through AI, \nelevate national defense strength and assure and protect national security.  \n\n\n8 \nIII.  Focus Tasks \n \nBased on the overall picture of national development, accurately grasp the global \ndevelopment trends of AI, find the correct openings for breakthroughs and directions for \nthe main thrust, comprehensively strengthen basic science and technology innovation \ncapabilities, comprehensively expand the depth and breadth of application in focus areas, \nand comprehensively enhance the built-in intelligence levels of applications in economic \nand social development, as well as in national defence. \n(1)  Build open and coordinated AI science and technology \ninnovation systems \n \nFocus on increasing the supply of AI innovation sources; strengthen deployments in areas \nsuch as advanced basic theory, key general technologies, basic platforms, talent teams, \netc.; stimulate open-source sharing; systematically enhance sustained innovation \ncapabilities; ensure that our country's AI science and technology levels ascend to the \nleading global ranks; and make ever more contributions to the development of global AI. \n1. Establish basic theory systems for a new generation of AI \n \nFocus on major advanced scientific AI questions; concurrently deal with present needs and \nlong-term developments; make breakthroughs in basic AI application theory bottlenecks; \ngive priority to deploying basic research that may trigger paradigmatic change in AI; \nstimulate the intersection and convergence of disciplines; and provide powerful scientific \nreserves for the sustained development and profound application of AI. \n \nMake breakthroughs in basic application theory bottlenecks. Aim at basic theoretical \norientations with clear applied objectives, which promise to trigger an upgrade of AI \ntechnology, strengthen basic theoretical research on big data intelligence, cross-media \nsensing and computing, human-machine blended intelligence, mass intelligence, \nautonomous cooperation and decision-making, etc. Focus on breakthroughs in big data \nintelligence, unsupervised learning, comprehensive deep reasoning and other such difficult \nissues. Establish data-driven cognitive computing models with natural language \nunderstanding at the core, and shape capabilities to go from big data to knowledge, and \nfrom knowledge to decision-making. Focus on breakthroughs in cross-media sensing and \ncomputing theory, including theories and methods for: low-cost and low-energy smart \nsensing, active sensing in complex landscapes, listening comprehension in the natural \nenvironment as well as language sensing, autonomous multimedia learning, etc. Realize \nsuperhuman sensing and highly-dynamic, high-dimensional, and multi-model distributed \nlarge-landscape sensing. The focuses on breakthroughs in blended and enhanced \nintelligence theory are: theories on human-machine cooperative and blended \nenvironmental understanding, decision-making, and learning; intuitive reasoning and \ncausal models, recall and knowledge evolution, etc.; realizing blended and enhanced \nintelligence where learning and reflection approach or exceed human intelligence levels. \nThe focuses for breakthroughs in collective intelligence theory are: theories and methods \nfor the organization, emergence and learning of collective intelligence; establishment of \nexpressible and computable mass intelligence incentive algorithms and models; and \nshaping Internet-based collective intelligence theory systems. The focuses for \n\n\n9 \nbreakthroughs in autonomous coordination, control and optimized decision-making theory \nare: theories concerning coordination sensing and interaction aimed at autonomous \nunmanned systems; autonomous coordination control and optimized decision-making; \nknowledge-driven human-machine-object triangular coordination and interoperation, etc.; \nand shaping novel theoretical systems and frameworks for innovation in autonomous \nintelligence and unmanned systems. \n \nArrange advanced basic theoretical research. Aim for a direction that may trigger a \nparadigmatic change in AI, far-sightedly arrange research on high-level machine learning, \nbrain-inspired intelligence computing, quantum smart computing, and other such cross-\ndomain basic theories. The focuses for breakthroughs in high-level machine learning \ntheory are theories and methods concerning self-adaptive learning, autonomous learning, \netc., and realizing AI with high interpretative and strong generalization capabilities. The \nfocuses for breakthroughs in brain-inspired intelligence computing theory are: theories \nconcerning brain-inspired information encoding, processing, recall, learning and reasoning; \nthe creation of brain-inspired complex systems and brain-inspired control theories and \nmethods; and establishment of new large-scale brain-inspired intelligence computing \nmodels and brain-inspired understanding computing models. The focuses for \nbreakthroughs in quantum computing theory are: methods for quantum-accelerated \nmachine learning; establishment of high-performance computing and quantum computing \nconvergence models; and shaping high-efficiency, accurate, and autonomous quantum AI \nsystem setups. \n \nLaunch cross-disciplinary exploratory research. Promote the intersection and convergence \nof AI with neurology, cognitive science, quantum science, psychology, mathematics, \neconomics, sociology and other such related basic disciplines; strengthen basic theoretical \nmathematical research to guide the development of AI algorithms and models; focus on \nresearching the basic theoretical questions of AI legal principles; support exploratory \nresearch that is strongly original, and where there is no consensus; encourage scientists to \nexplore freely; dare to overcome front-line scientific difficulties in AI; create ever more \noriginal theory; and make ever more original discoveries. \n \nBox 1: Basic Theories \n1. Big data intelligence theory. Research new data-driven and knowledge-driven AI \nmethods, theories and methods for sensing computing theory with natural language \nunderstanding, images and figures at the core, comprehensive deep reasoning and \ncreative AI theories and methods, basic theories and frameworks on smart decision-\nmaking with incomplete information, data-driven common AI data models and \ntheories, etc. \n2. Cross-media sensing and computing theory. Research sensing that exceeds human \nvisual abilities, active visual sensing and computing aimed at the real world, \nauditory sensing and computing of natural acoustic scenes, language sensing and \ncomputing in an environment of natural interaction, human sensing and computing \naimed at asynchronous orders, autonomous learning aimed at smart media sensing, \nand urban omnidimensional smart sensing and reasoning engines. \n3. Hybrid and enhanced intelligence theory. Research hybridization and convergence \nwhere “the human is in the loop,” behavioral strengthening through human-\nmachine smart symbiosis and brain-machine coordination, intuitive machine \n\n\n10 \nreasoning and causal models, associative recall models and knowledge evolution \nmethods, complex data and task blended and enhanced intelligence learning \nmethods, cloud robotics coordination computing methods, and situational \ncomprehension and human-machine group coordination in real-world \nenvironments. \n4. Swarm intelligence theory. Research swarm intelligence structural theory and \norganizational methods, swarm intelligence incentive mechanisms and emergence \nmechanisms, swarm intelligence learning theories and methods, common swarm \nintelligence computing paradigms and models. \n5. Autonomous coordination and control, and optimized decision-making theory. \nResearch coordination sensing and interaction aimed at autonomous unmanned \nsystems, coordination, control and optimized decision-making aimed at \nautonomous and unmanned systems, knowledge-driven human-machine-object \ntriangular coordination and interoperability theories. \n6. High-level machine learning theory. Research basic statistical learning theories, \nreasoning and decision-making under uncertainty, distributed learning and \ninteraction, learning while protecting privacy, small-sample learning, deep intensive \nlearning, unsupervised learning, semi-supervised learning, active learning and other \nsuch learning theories and efficient models. \n7. Brain-inspired intelligence computing theory. Research theories and methods on \nbrain-inspired sensing, brain-inspired learning, and brain-inspired recall \nmechanisms and computing blends, brain-inspired complex systems, brain-inspired \ncontrol, etc. \n8. Quantum intelligent computing theory. Explore cognitive quantum models and \nintrinsic mechanisms, research efficient quantum intelligence models and \nalgorithms, high-performance and high-bitrate quantum AI processors, real-time \nquantum AI systems that can exchange information with the outside world, etc. \n \n2.  Build a next-generation AI key general technology system \n \nFocusing on the urgent need to raise China's international competitiveness in AI, next-\ngeneration AI key general technology R&D and deployment should make algorithms the \ncore; data and hardware the foundation; and upping capabilities in sensing and \nrecognition, knowledge computing, cognitive reasoning, executing motion, and human-\nmachine interface the emphasis; in order to form openly compatible, stable and mature \ntechnological systems. \n \nKnowledge computing engine and knowledge service technology. Key breakthroughs in \nknowledge processing, deep search, and visual interactive core technology; realization of \nautomatic acquisition of incrementally growing knowledge; possession of concept \ndiscernment, object discovery, attribute prediction, evolutionary knowledge modeling, and \nrelationship discovery capabilities; the formation of multi-billion-scale, multi-source, \nmulti-disciplinary, multi-data type, and cross-medium knowledge maps. \n \nCross-medium analytical reasoning technology. Key breakthroughs in cross-medium \nunified indicators; relational understanding and knowledge mining; knowledge map \nstructure and learning; knowledge evolution and reasoning; intelligent description and \n\n\n11 \ngeneration, etc., technology. Realization of cross-medium knowledge indicators, analysis, \nmining, reasoning, evolution, and utilization. Construct analytic reasoning engines. \n \nKey swarm intelligence technology. Key breakthroughs on the basis of the popularization \nof the internet, mass collaboration, knowledge resource management, and open sharing, \netc., technologies. Building frameworks to display swarm intelligence knowledge. Realize \nthe integration and strengthening of swarm intelligence-based knowledge acquisition and \nswarm intelligence under open development conditions. Support swarm perception, \ncooperation, and evolution at a national, tens-of-millions scale.  \n \nNew architecture and new technology for hybrid and enhanced intelligence. Key \nbreakthroughs in human-machine interaction for perception and execution integration \nmodels, new types of intelligent computing-fronted sensors, common use hybrid \narchitecture, etc., core technologies. Build autonomous, environmentally adaptable hybrid \nenhanced intelligent systems, human-machine hybrid enhanced intelligent systems and \nsupport environments. \n \nIntelligent technologies of autonomous unmanned systems. Key breakthroughs in \nautonomous unmanned system computing architecture, complex situational environment \nperception and understanding, real-time accurate positioning, adaptable, intelligent \nnavigation in complex environments, etc., general technologies. Unmanned and \nautonomously controlled systems including automobiles, ships, automatic driving in \ntraffic, etc., intelligent technologies. Develop service robots, special-purpose robots, etc., \ncore technologies and support unmanned system application and manufacturing \ndevelopment.  \n \nIntelligent virtual reality modeling technology. Key breakthroughs in intelligent modeling \ntechnology for virtual counterparts. Increasing the sociality, diversity, and lifelike quality of \nvirtual reality intelligent counterpart behavior. Realize the organic integration, high \nefficiency, and interactivity of virtual reality and augmented reality, etc., technologies. \n \nIntelligent computing chips and systems. Key breakthroughs in high energy \nefficiency, reconfigurable brain-inspired computing chips and brain-inspired visual sensor \nsystems with computational imaging capabilities. Research and develop high-efficiency \nbrain-inspired neural network architectures and hardware systems with autonomous \nlearning capabilities. Realize brain-inspired intelligent systems with multimedia sensory \ninformation understanding, intelligence growth, and common sense reasoning capabilities. \n \nNatural language processing technology. Key breakthroughs in natural language grammar \nlogic, word-concept symbols, and deep semantic analysis core technologies. Advance \neffective human-machine communication and free interaction. Realize multi-style, multi-\nlanguage, multi-domain natural language intelligent understanding and automated \n[results] generation. \n \n \n \n\n\n12 \nBox 2: Key General Technologies \n \n1. Knowledge computing engines and knowledge service technology. Researching \nknowledge computing and visual interaction engines; researching innovative \ndesign, digital creation, and commercial intelligence with visual media at the core; \ndeveloping large-scale organic data knowledge discovery. \n2. Cross-medium analytic reasoning technology. Researching cross-medium unified \nindicators, connected understanding and knowledge mining, knowledge map \nbuilding and learning, knowledge evolution and inference, intelligent description \nand generation, etc., technology; developing cross-medium analytic reasoning \nengine and verification systems. \n3. Key swarm intelligence technology. Developing swarm intelligence's active \nperception and discovery, knowledge gain and generation, cooperation and sharing, \nevaluation and evolution, human-machine integration and enhancement, self-\npreservation and mutual security, etc., key technology studies; building service \nsystem architecture for the crowd intelligence space; researching mobile crowd \nintelligent coordinated decision making and control technologies. \n4. Hybrid enhanced intelligent new architectures and technologies. Researching hybrid \nenhanced intelligent core technology and cognitive computing frameworks; new-\nmodel hybrid computing architectures, human-machine collective driving, online \nintelligent learning technology, and hybrid enhanced frameworks for simultaneous \nmanagement and control. \n5. Autonomous unmanned systems intelligent technology. Researching unmanned \nautonomous control intelligent technology for automobiles, ships, traffic, automatic \ndriving, etc.; service, space, maritime, and polar robot technology; unmanned \nworkshop/intelligent factory intelligent technology; high-end intelligent control \ntechnology and autonomous unmanned operating systems. Researching \npositioning, navigation, recognition, etc., robotic and mechanical arm autonomous \ncontrol technology for visual sensing in complex environments.  \n6. Virtual reality intelligent modeling technology. Researching mathematical \nexpression and modeling methods for virtual counterpart intelligent behavior; \nproblems such as natural, persistent, and deep exchange between users and virtual \ncounterparts and virtual environments; intelligent counterpart modeling technology \nand method systems. \n7. Intelligent computing chips and systems. Researching neural network processors, \nas well as high-energy efficiency, reconfigurable brain-inspired computing chips, \netc.; new-model perception chips and systems, intelligent computing system \nstructure and systems, and AI operating systems. Researching architectures \nsuitable for AI hybrid architectures, etc. \n8. Natural language processing technology. Researching short text computing and \nanalysis technology, cross-language text mining technology and turning toward \nsemantic comprehension technology for machine cognitive intelligence, and \nhuman-machine interaction systems for multimedia information comprehension. \n \n \n \n \n \n \n\n\n13 \n3. Coordinate the layout of AI innovation platforms \n  \nConstruct AI innovation platforms. Strengthen the foundational support for AI research and \ndevelopment and applications. AI open-source hardware and software infrastructure \nplatforms should focus on building and supporting unified computing frameworks for \nknowledge reasoning, probability statistics, depth learning, and other AI paradigms. Form \nand promote an ecological chain of platforms for interaction and synergies among AI \nsoftware, hardware, and intelligent clouds. The group intelligent service platform should \nfocus on the construction of knowledge resource management and the open sharing tools \nbased on the large-scale cooperation on the Internet. Create a platform and service \nenvironment for the innovation of the industry and university. The hybrid enhanced \nintelligent support platforms should focus on the construction of a heterogeneous real-\ntime computing engine supporting large-scale training and a new computing clusters, \nproviding a service-oriented, systematic platform and solution for complex intelligent \ncomputing. Autonomous unmanned system support platform focuses on the construction \nof autonomous system environmental awareness, autonomous collaborative control, \nintelligent decision-making and other AI common core technology support systems. Create \ndevelopment and test environments for open, modular, reconfigurable autonomous \nunmanned systems. AI basic data and security detection platforms should focus on the \nconstruction of AI for the public data resource library, the standard test data set, cloud \nservice platform, the formation of AI algorithms and platform security test evaluation \nmethods, techniques, norms and tools, promoting the open sourcing and openness of all \nkinds of common software and technology platform. Promote military-civilian sharing and \njoint use for all kinds of platforms in accordance with the requirements of deep military-\ncivil integration related provisions. \n \nBox 3: Basic Support Platforms \n \n1. AI Open-Source Hardware and Software Infrastructure and Platforms. Establish big \ndata and AI open-source software platforms, terminal, and cloud collaborative AI \ncloud service platforms, new multi-intelligent sensor and integrated platforms, new \nproduct design platforms based on AI hardware, and future network, big data \nintelligent service platforms. \n2. Group Intelligent Service Platforms. Establish group knowledge-based computing \nand support platforms, science and technology public service systems, group \nintelligent software development and verification automation systems, group \nintelligent software learning and innovation systems, open environment cluster \ndecision-making systems, and group-sharing economic service systems. \n3. Hybrid Enhanced Intelligent Support Platforms. Establish AI supercomputing \ncenters, large-scale super intelligent computing support environments, online \nintelligent education platforms, “human-in-the-loop” driving brains, intelligent \nplatforms for complexity analyses and risk assessment in industrial development, \nintelligent security platforms to support nuclear power security operations, and \nresearch and development and testing platforms for human-machine joint driving \ntechnology. \n4. Autonomous Unmanned System Support Platforms. Establish common core \ntechnology and support platforms, independent unmanned systems, independent \ncontrol of unmanned aerial vehicles, and automatic driving support platforms for \nauto, ship and rail traffic, service robots, space robots, marine robots, polar robot \n\n\n14 \nsupport platforms, technical support platforms for intelligent factory and intelligent \ncontrol equipment, etc. \n5. AI Basic Data and Security Detection Platforms. Construct artificial data-oriented \npublic data resource libraries, standard test data sets, and cloud service platforms. \nEstablish test models and evaluation models for the security of AI algorithms and \nplatforms. Research and develop security evaluation tools for AI algorithms and \nplatforms. \n4. Accelerate the training and gathering of high-end AI talent \nMake the construction of a high-end talent team of the utmost importance in the \ndevelopment of AI. Adhere to the combination of training and introduction. Improve the AI \neducation system, strengthen the construction of a talent pool and echelons, especially \naccelerate the introduction of the world’s top talent and young talent, forming China’s AI \ntop talent base. \nCultivate high-level of AI innovative talents and teams. Support and cultivate the \ndevelopment potential of leading AI talent. Strengthen professional and technical \npersonnel training for basic research, applied research, operations and maintenance \naspects of AI. Pay attention to the training of compound talents, focusing on cultivating \nvertical composite talents for AI theory, methods, technology, products, and application, \nand compound talents who master the “AI +” economy, society, management, standards, \nlaw, and other horizontal areas. Through major research and development tasks and base \nand platform construction, converge high-end talents in AI. Create high-level innovation \nteams in a number of AI key domains. Encourage and guide domestic innovative talents \nand the teams to strengthen cooperation with the world’s top AI research institutions. \n \nIncrease the introduction of high-end AI talent. Open up specialized channels and \nimplement special policies to achieve the precise introduction of peak AI talent. Focus on \nthe introduction of international top scientists and high-level innovation teams in neural \nawareness, machine learning, automatic driving, intelligent robots, and other areas. \nEncourage the use of flexible introduction of AI talent through project cooperation, \ntechnical advice, etc. Coordinate the use of the “Thousands Talents” plan and other \nexisting talent plans to strengthen the field of AI talents, especially through the \nintroduction of outstanding young talent. Improve enterprise human capital cost \naccounting and related policies. Encourage enterprises and scientific research institutions \nto introduce AI talent. \n \nConstruct an AI academic discipline. Improve the disciplinary layout of the AI domain. \nEstablish AI majors. Promote the construction of a discipline in the domain of AI. Establish \nAI institutes as soon as possible in pilot institutions. Increase the enrollment places for \nmasters and PhDs in working in AI and related disciplines. Encourage colleges and \nuniversities to broaden the content of AI professional education on an original basis. Create \na new model of “AI + X” compound professional training, attaching importance to cross-\nintegration of professional education for AI and mathematics, computer science, physics, \nbiology, psychology, sociology, law, and other disciplines. Strengthen cooperation in \nproduction and research. Encourage universities, research institutes, enterprises and other \ninstitutions to carry out the construction of an AI discipline. \n \n\n\n15 \n(2) Fostering a high-end, highly efficient smart economy \n \nAccelerate the fostering of an AI industry with a major leading and driving effect, stimulate \nthe profound convergence of AI and all industrial areas, and create data-driven smart \neconomic patterns with human-machine coordination, cross-sectoral convergence, and \njoint creation and sharing. Data and knowledge will become the first factor for economic \ngrowth; human-machine coordination will become the mainstream method of production \nand service; cross-sectoral convergence will become an important economic model; joint \ncreation and sharing will become basic characteristics of the economic ecology; \nindividualized demands and made-to-order will become new consumption trends; and \nproductivity will increase substantially, drive industries to migrate towards the high end of \nvalue chains, powerfully support the development of the real economy, and \ncomprehensively increase the quality and efficiency of economic development. \n \n1. Forcefully develop new AI industries \n \nAccelerate the transformation and application of key AI technologies, stimulate the \nintegration of technologies with commercial model innovation, promote the innovation of \nsmart products in focus areas, vigorously foster new AI business models, compose high-\nend industry chains, and forge AI industry groups with international competitiveness. \n \nSmart software and hardware. Develop operating systems, databases, intermediary \ndevices, development tools, and other such key software and hardware aimed at AI; make \nbreakthroughs in graphic processing and other such core hardware; research solution \nplans for smart systems in pattern recognition, voice understanding, machine translation, \nsmart interaction, knowledge processing, control and decision-making, etc.; and foster \nand expand basic software and hardware industries aimed at AI. \n \nSmart robots. Tackle core components and special sensors for smart robots, perfect \nhardware interface standards, software interface standards, and safe usage standards for \nsmart robots. Research and develop smart industrial robots and smart service robots, \nrealize large-scale application, and enter into global markets. Research, produce, and \npopularize space robots, maritime robots, polar robots, and other such special kinds of \nsmart robots. Establish smart robot standard systems and security norms. \n \nSmart delivery tools. Develop self-driving vehicles and rail traffic systems; strengthen the \nintegration and coordination of vehicle load sensing, automatic driving, the Internet of cars, \nthe Internet of Things, and other such technologies; develop smart traffic sensing systems, \ncreate national indigenous automatic driving platform technology systems and industrial \nassembly capabilities; and explore self-driving vehicle sharing models. Develop consumer \nand commercial unmanned aircraft and unmanned ships, and establish and trial \nspecialized service systems for authentication, monitoring, technology competition, etc., \nperfect management measures for the space and maritime areas. \n \nVirtual reality and augmented reality. Make breakthroughs in key technologies such as \nhigh-performance software modelling, content capturing and generation, augmented \nreality and human-machine interaction, integrated environments and tools, etc. Research \nand create virtual display devices, optical devices, high-performance three-dimensional \ndisplay devices, development engines, and other such products. Establish standards and \n\n\n16 \nevaluation systems for virtual reality and augmented reality technologies, products, and \nservices, and promote their converged application in focus sectors. \n \nSmart terminals. Accelerate the research and development of smart terminal core \ntechnologies and products, develop new-generation smart phones, on-board smart \nterminals for cars, and other such mobile smart terminal products and equipment. \nEncourage the research and development of smart watches, smart earpieces, smart \nglasses, and other such wearable terminal products, and expand product forms and \napplication services. \n \nBasic Internet of Things devices. Develop high-sensitivity and highly reliable smart sensors \nand chips supporting the new-generation Internet of Things. Make progress in core Internet \nof Things technologies such as RFID and short-distance machine communications, as well \nas key components such as low-power processors. \n \n2. Accelerate and promote the upgrade of industrial intelligentization \n \nPromote the converged innovation of AI in all sectors. Launch AI application \ndemonstrations and trials in focus sectors and areas such as manufacturing, agriculture, \nlogistics, finance, commerce, household goods, etc. Promote the application of AI at scale, \nand comprehensively upgrade the smartness level of industrial development. \n \nSmart manufacturing. Focus on the major demands for building a strong manufacturing \ncountry, move forward the integrated application of systems such as key technologies and \nequipment for smart manufacturing, core supporting software, the industrial internet, etc. \nResearch and develop smart products and smart connected products, tools and systems \nthat can be used in smart manufacturing, and smart manufacturing cloud service \nplatforms. Popularize smart manufacturing processes, distributed smart manufacturing, \nnetworked coordinated manufacturing, long-distance diagnosis and operational services, \nand other such novel manufacturing models. Establish smart manufacturing standard \nsystems, and move forward with the intelligentization of manufacturing activities across \nthe entire lifecycle. \n \nSmart agriculture. Research and formulate smart agricultural sensing and control systems, \nsmart agricultural equipment, autonomous tasking systems for farming equipment across \nfields, etc. Establish and complete smart agriculture information remote sensing and \nmonitoring networks integrating air, space, and land components. Establish model \nagriculture big data smart decision-making and analysis systems, launch trials of smart \nfarms, smart plant factories, smart pastures, smart fisheries, smart orchards, smart farm \nproduce processing workshops, green and smart farm product supply chains and other \nsuch integrated applications. \n \nSmart logistics. Strengthen research, development and broad use of smart logistics \nequipment for smart loading, unloading, and transportation; parcel sorting, processing and \ndelivery; etc. Establish smart deep-sensing storage systems, and enhance storage and \noperational management levels and efficiency. Perfect smart logistics public information \nplatforms and command systems, product quality authentication and tracing systems, \nsmart distribution and dispatch systems, etc. \n \n\n\n17 \nSmart finance. Establish big data systems for finance, and enhance multimedia data \nprocessing and comprehension capabilities for finance. Innovate smart financial products \nand services, develop new financial business models. Encourage the financial sector to use \nsmart customer service, smart inspection, and other such technologies and equipment. \nBuild smart warning and prevention systems for financial risk. \n \nSmart commerce. Encourage the application of cross-media analysis and reasoning, \nknowledge computing engines and knowledge services, and other such new technologies \nin the commercial area, and popularize AI-based novel commercial services and decision-\nmaking systems. Build cross-medium data platforms covering geographic positioning, \nonline media, urban basic data, etc., and support enterprises' launching smart services. \nEncourage the provision of made-to-order commercial smart decision-making services \nfocusing on individual demands and enterprise management. \n \nSmart household goods. Strengthen the converged application of AI technology and \nhousehold and building systems, and enhance the smartness levels of building facilities \nand household goods. Research, develop, and use household connection and interactivity \nagreements, as well as interface standards suited for different application settings. \nEnhance sensing and connection capabilities of household electrical appliances, durable \ngoods and other such household products. Support smart household enterprises in \ninnovating new service models, and promote interactive and sharing solutions and plans. \n \n3. Forcefully develop smart enterprises \n \nPromote the upgrading of enterprises' smartness levels on a large scale. Support and guide \nenterprises to use new AI technologies in core operational segments such as design, \nproduction, management, logistics, sales, etc. Build novel enterprise organization \nstructures and operational models; create smart and converged business models for \nmanufacturing, services, and finance; and develop individualized made-to-order; and \nbroaden smart product supply. Encourage large-scale Internet enterprises to build cloud \nmanufacturing platforms and service platforms, and provide online key industry software \nand model databases aimed at manufacturing enterprises. Launch outsourcing services for \nmanufacturing capacity, and promote the development of smartness among small and \nmid-size enterprises. \n \nPopularize the use of smart factories. Strengthen the application and demonstration of key \ntechnologies and system methods for smart factories. Focus on popularizing production \nline reconstruction and dynamic smart control, production faculty smart interconnection \nand cloud data collection, multi-dimensional human-machine-object coordination, \ninteroperability, and other such technologies. Encourage and guide enterprises to build \nfactory big data systems, networked distributed production facilities, etc. Realize the \nnetworking of production equipment, the visualization of production data, the transparency \nof production processes, and the automation of production sites; and enhance the \nsmartness levels of factory operational management. \n \nAccelerate the fostering of AI industry-leading enterprises. Accelerate the creation of \nglobal leading AI enterprises and brands in advantageous areas such as unmanned aircraft, \nspeech recognition, pattern recognition, etc. Accelerate the fostering of a batch of key \nenterprises in novel areas such as smart robots, smart cars, wearable equipment, virtual \nreality, etc. Support AI enterprises to strengthen their patent structures, and take the lead \n\n\n18 \nin or participate in the formulation of international standards. Promote domestic \nadvantageous enterprises, sectoral organizations, scientific research bodies, higher \neducation institutes, etc., to jointly establish the AI Industry and Technology Innovation \nAlliance of China. Support key backbone enterprises to build open source hardware \nfactories, open source software platforms, create innovative ecologies integrating all kinds \nof resources, stimulate small and mid-size AI enterprises to develop and to be used in all \nareas. Support all kinds of bodies and platforms to provide specialized services aimed at AI \nenterprises. \n4. Create AI innovation heights \nCombined with each locality’s foundation and advantages, according to the field of AI \napplications classifications, advance the layout of the relevant industries. Encourage local \nindustry chains and innovation chains around AI. Gather high-end factors, high-end \nenterprises, and high-end talent. Build AI industry clusters and heights of innovation. \n \nLaunch AI innovation application pilot demonstrations. In areas where the AI foundation is \nfavorable and its development potential bigger, organize and launch national AI innovation \nexperiments. Explore systems and mechanisms, policy and regulation, the cultivation of \ntalent, and other major reforms. Promote the transformation of the AI achievements, major \nproduct integrated innovation, and demonstration of applications. Form replicable, \npromotable experience, leading to the promotion of intelligent economy and intelligent \nsocial development. \n \nConstruct national AI industrial parks. Rely upon national independent innovation \ndemonstration areas and the national high-tech industry development zone and other \ninnovative vectors. Strengthen science and technology talent, finance, policy, and other \nelements of the optimal allocation and combination. Accelerate the construction of AI \nindustry innovation cluster. \n \nConstruct national AI mass innovation bases. Relying on colleges and universities and \nscientific research institutes concentrated in localities, build AI field professionalized \ninnovation platforms and other new entrepreneurial service agencies. Construct a number \nof low-cost, convenient, all-factor, open-style AI ‘hackerspaces.’ Improve incubation \nservices system, promote the transformation of AI scientific and technological \nachievements, and support AI innovation and entrepreneurship. \n(3) Construct a safe and convenient intelligent society \n \nBased on the goal of improving people's living standards and quality, speed up and deepen \nthe applications of AI, increase the level of intelligentization of the whole society to form an \nall-encompassing and ubiquitous intelligent environment. Increasingly, repetitive, \ndangerous tasks will be completed by AI, while individual creativity will play a greater role. \nForm more high-quality and high comfort jobs; make precision intelligent services more \ndiverse, such that people can maximize their enjoyment of high quality services and \nconvenient life. Through a substantial increase in the level of intelligentization of social \ngovernance, make social operations more safe and efficient. \n\n\n19 \n1. Develop convenient and efficient intelligent services \n \nAccelerate the application of innovative AI throughout education, health care, pension and \nother urgent needs involving people's livelihood, to provide for the public personalized, \ndiversified, high-quality services. \n \nIntelligent Education. Utilize intelligent technology to accelerate and promote a personnel \ntraining model and reform to teaching methods; establish new-type education systems, \nincluding intelligent learning and interactive learning. Launch the construction of \nintelligent campuses; promote AI in teaching, management, resource construction, and \nother full-scale applications. Develop three-dimensional integrated teaching field, based \non big data intelligent online learning and education platforms. Develop intelligent \neducational assistants; establish intelligent, fast and comprehensive education analysis \nsystem. Establish a learner-centered educational environment, and provide precision-\ndeployed education services, achieve daily education and lifelong education. \n \nIntelligent Medical Care. Promote the use of new models and new methods of AI treatment, \nestablish a rapid, accurate intelligent medical system. Explore intelligent hospital \nconstruction, develop human-machine coordinated surgical robots and intelligent clinic \nassistants. Pursue research and development on flexible wearable, biologically compatible \nphysiological monitoring systems, research and development of human-computer \ncollaboration intelligent clinical diagnosis and treatment programs. Achieve intelligent \nimage recognition, pathology classification, and intelligent multi-disciplinary consultation. \nCarry out large-scale genome recognition, proteomics, metabolomics, and other research \nand development of new drugs based on AI, promote intelligent pharmaceutical regulation. \nStrengthen epidemic intelligence monitoring, prevention, and control. \n \nIntelligent Health and Elder Care Systems. Strengthen community intelligent health \nmanagement, achieve breakthroughs in big data analysis, Internet of Things, and other key \ntechnologies. Research and develop health management wearable equipment and home \nintelligent health testing and monitoring equipment. Promote changes in health \nmanagement from point-like monitoring to continuous monitoring, from short process \nmanagement to long process management. Construct intelligent elder care communities \nand institutions; build a safe and convenient intelligent pension infrastructure system. \nStrengthen the intelligentization of products for elderly persons and intelligent products \nsuitable for the aged. Develop audio-visual aid equipment, physical auxiliary equipment, \nand other intelligent home care equipment, expanding the elderly’s activity space. Develop \nmobile social and service platform for the elderly and emotional escort assistant to \nenhance the quality of life of the elderly. \n2. Promote the intelligentization of social governance \n \nPromote the application of AI technology for administrative management, judicial \nmanagement, urban management, environmental protection, and other hot and difficult \nissues in social governance, to promote the modernization of social governance. \n \nIntelligent Government. Develop an AI platform for government services and decision-\nmaking. Develop a decision-making engine for the open environment. Promote \napplications in research on complex social problems, policy assessment, risk warning, \n\n\n20 \nemergency response, and other major matters of strategic decision-making. Strengthen \nthe integration of government information resources and accurate forecasting of public \ndemands, and smooth communication channels between the government and the public. \n \nSmart Courts. Construct a set of trial, personnel, data applications, judicial disclosure, and \ndynamic monitoring into an integrated court data platform. Promote AI applications for \napplications including evidence collection, case analysis, and legal document reading and \nanalysis. Achieve the intelligentization of courts and trial systems and trial capacity.  \n \nSmart Cities. Build an intelligentized city infrastructure, develop intelligent buildings, and \npromote the intelligentization, transformation, and upgrading of underground corridors and \nother municipal infrastructure. Construct urban big data platforms to build a \nheterogeneous, integrated data system for urban operations and management. Achieve \ncomprehensive perception and deep understanding of the operation of complex urban \nsystems for urban infrastructure and urban green space, wetlands, and other important \necological elements. Research and develop to build community public service information \nsystems. Promote community service system and residents’ intelligent home system \ncollaboration. Promote the intelligentization of the full lifecycle of urban planning, \nconstruction, and management. \n \nSmart Transportation. Research, establish, and operate vehicle automatic driving and road \ncoordination technology systems. Research and develop information and integrated data \nplatforms for transportation under complex multi-dimensional conditions. Establish \nintelligentized transportation command, control, and integrated operations. Actualize \nintelligent transportation obstacle removal and integrated management and coordination \nand command. Build intelligent transportation monitoring, management, and service \nsystems covering the ground, tracks, low altitude, and the sea. \n \nIntelligent Environmental Protection. Establish an intelligent monitoring large data \nplatforms and systems covering the atmosphere, water, soil, and other environmental \nareas. Build information-sharing and intelligent environmental monitoring networks and \nservice platforms for coordination of land and sea, integration of atmosphere and earth, \nand upwards and downwards synergies. Research and develop intelligent forecasting \nmodels and method and early warning programs for energy resource consumption and \nenvironmental pollutant discharge. Strengthen the Beijing-Tianjin-Hebei, Yangtze River \nEconomic Zone, and other major national strategic regions’ construction of intelligent \nprevention and control system for environmental protection and sudden environmental \nevents. \n3. Use AI to enhance public safety and security capabilities \nAdvance the deepening of AI applications in the field of public safety. Promote the \nconstruction of public safety and intelligent monitoring and early warning and control \nsystems. Research and develop a variety of detection sensor technology, video image \ninformation analysis and identification technology, biometric identification technology, \nintelligent security and police products. Establish intelligent monitoring platform for \ncomprehensive community management, new criminal investigations, anti-terrorism, and \nother urgent needs. Strengthen the upgrading and intelligentization of security equipment \nfor key public areas. Support carrying out public security regional demonstrations based on \nAI according to the conditions of the community or the city. Strengthen the use of AI for \nfood safety protection, food classification, warning level, food safety risks and assessment, \n\n\n21 \nand the establishment of intelligent food safety early warning system. Strengthen the \neffective monitoring of natural disasters, natural disasters, around the earthquake disaster, \ngeological disasters, meteorological disasters, floods and disasters and marine disasters \nand other major natural disasters, to build an intelligent monitoring and early warning and \ncomprehensive response platform. \n4. Promote social interaction and mutual trust \nGive full play to the role of AI technology in enhancing social interaction and promoting \ncredible communication. Strengthen the next generation of social network research and \ndevelopment, accelerate innovation in augmented reality, virtual reality, and other \ntechnologies to promote the integrative use of virtual environments and physical \nenvironments to meet personal perception, analysis, judgment and decision-making real-\ntime information needs, and to achieve the smooth transition of different scenes of work, \nstudy, life, and entertainment. In order to improve the interpersonal communication needs, \ndevelop intelligent assistant products with the ability to accurately understand the needs \nof emotional interaction. Promote the integration of blockchain technology and AI, \nestablish a new social credit system, and minimize the cost and risks of interpersonal \ncommunication. \n(4) Strengthen military-civilian integration in the AI domain \n  \nDeepen implementation of military-civilian integration development strategy, to promote \nthe formation of an all-element, multi-field, high efficiency AI military-civilian integration \npattern. Build new generation AI based on research and development in the common \ntheory and critical common technology. Establish mechanisms to normalize \ncommunication and coordination among scientific research institutes, universities, \nenterprises and military industry units. Promote military-civilian two-way transformation of \nAI technology. Strengthen a new generation of AI technology as a strong support to \ncommand and decision-making, military deduction, defense equipment, and other \napplications. Guide defense domain AI technology toward civilian applications. Encourage \nand advantage people’s scientific research forces to participate in the domain of national \ndefense for major scientific and technological innovation tasks in AI. Promote all kinds of AI \ntechnology to become quickly embedded in the field of national defense innovation. \nStrengthen the construction of military and civilian AI technology standard systems.  \nPromote the overall layout and open sharing of science and technology innovation \nplatforms and bases. \n(5) Build a safe and efficient intelligent infrastructure system \n \nVigorously promote the construction of intelligent information infrastructure. Enhance the \ntraditional level of intelligent infrastructure to form a smart economy, intelligent society \nand national defense needs of the infrastructure system. Speed up the promotion of \ninformation transmission as the core of the digital, network information infrastructure. \nTake integration awareness, transmission, storage, computing, and processing in \nintelligent information infrastructure changes. Optimize network infrastructure, research \nand develop the layout of fifth generation mobile communication (5G) systems. Improve the \nInternet of Things infrastructure. Accelerate the integration of information network \nconstruction. Improve low-latency, high-throughput transmission capacity. Coordinate the \n\n\n22 \nuse of big data infrastructure, strengthen data security and privacy protection, to provide \nmassive data support for AI research and development and extensive applications. Build \nhigh-performance computing infrastructure, and enhance the service support capabilities \nof supercomputing centers for AI applications. Construct distributed and efficient energy \nInternet, form multi-energy support complementary, timely, and effective access to new \nenergy networks. Promote intelligent energy storage facilities, intelligent electricity \nfacilities, energy supply and demand information to achieve real-time matching and \nintelligent response. \n  \n  \nBox 4: Intelligentized Infrastructure \n \n1.   Network Infrastructure. Speed up the layout of real-time collaborative AI 5G \nenhanced technology research and the development and application of space-\noriented collaborative AI for the construction of high-precision navigation and \npositioning networks to strengthen the core of intelligent sensing technology \nresearch and key facilities. Develop intelligent industrial support, driving networks, \netc., to study the intelligent network security architecture. Speed up the \nconstruction of integrated information network for space and earth, promoting a \nspace-based information network, the future of the Internet, mobile communication \nnetwork of the full integration. \n2. \n   Big Data Infrastructure. Rely on a national data sharing exchange platform, open \ndata platform and other public infrastructure. Construct governance, public \nservices, industrial development, technology research and development, and other \nfields of big data information databases Support the implementation of national \ngovernance data applications. Integrate various types of social data platforms and \ndata center resources. Create nationwide integrated service capabilities with \nreasonable layout and linkages. \n3. High-performance computing infrastructure. Continue to strengthen the \nsupercomputing infrastructure, distributed computing infrastructure and cloud \ncomputing center construction. Build sustainable development of high-\nperformance computing application for the ecological environment. Promote the \nnext generation of supercomputer research and development and applications. \n(6) Plan a new generation of AI major science and technology \nprojects \n \nFor the development of China’s AI needs and weak links, establish of a new generation of AI \nmajor scientific and technological projects. Strengthen the overall co-ordination, clear the \nboundaries of the tasks and the focus of research and development. Form a new \ngeneration of AI major scientific and technological projects as the core, and use existing \nR&D layout to support the “1 + N” AI program. \n \n“1” refers to a new generation of AI scientific and technological mega-projects, focusing on \nforward-looking layout for basic theories and key common technologies, including the \nstudy of big data intelligence, cross-media perception and computing, hybrid enhanced \nintelligence, group intelligence, autonomous collaborative control, and decision-making \ntheory. Research knowledge computing engines and knowledge service technologies, \ncross-medium analysis reasoning technology, key swarm intelligence technologies, new \n\n\n23 \narchitecture and new technology for hybrid enhanced intelligent, autonomous unmanned \ncontrol technology, and basic theory and common technology for open-source shared AI. \nContinue to carry out the development of AI prediction and research, strengthening the \neconomic and social impact of and countermeasures for AI. \n \n“N” refers to the national planning and deployment of AI research and development \nprojects. Focusing on strengthening the new generation of AI with the convergence major \nscientific and technological projects, collaborative impetus for research, technological \nbreakthroughs and product development applications. Strengthen the convergence of \nmajor national science and technology projects. Support AI hardware and software \ndevelopment in the “Hegaoji” Megaproject,1 integrated circuit equipment and other national \nscience and technology major projects. Strengthen mutual support for AI and other \n“Technological Innovation 2030 - Mega-Projects.” Accelerate the use of AI to provide \nsupport for major technical breakthroughs in brain science and brain computing, quantum \ninformation and quantum computing, intelligent manufacturing and robotics, and big data \nresearch. The National Key Research and Development Plan will continue to promote high-\nperformance computing and other key special applications, while increasing support for AI-\nrelated technology research and development and application; the National Natural \nScience Foundation will strengthen cross-disciplinary research and support for free \nexploration in the field of AI. Focus on special deployment and strengthen the application \nof AI technology demonstrations to the deep sea space station, health protection, and \nother major projects, smart cities, intelligent agricultural equipment and other Key National \nR&D Projects. Support the openness and sharing of research results on basic theory of AI \nand common technology through other basic science and technology plans. \n \nInnovate in the organization and implementation of models for new generation AI major \nscientific and technological projects. Adhere to focus on doing things, focusing on the \nprinciple of breakthrough. Give full play to the role of market mechanisms to mobilize \ndepartments, local, business and social forces to promote the implementation of all \naspects. Pursue clear management responsibility, regular assessments, to strengthen the \ndynamic adjustments and improve management efficiency. \nIV.  Resource Allocation \n \nFully use existing finances, bases and other such stored resources, comprehensively plan \nthe allocation of international and domestic innovation resources, give rein to the guiding \nrole of finance administration input and policy incentives, and the dominant role of the \nmarket in allocating resources, impel enterprises and society to expand input, and create a \nnew pattern of multi-sided support through finance administration funding, financial \ncapital, and social capital. \n                                               \n \n1 Translator’s note: This refers to the Medium and Long-term Plan for S&T Development \n2006-2020 megaproject: core (he) electronic devices, high-end (gao) general-purpose \nchips, and basic (ji) software. \n\n\n24 \n(1) Establish financial support mechanisms guided by the \nfinancial administration and dominated by the market \n \nComprehensively plan multiple-channel financial input by government and markets, \nstrengthen support through finance administration funding, enliven existing resources, and \nprovide support for fundamental and advanced AI research, critical public technology \nbreakthroughs, result transformation, base and platform construction, innovative \napplication demonstrations, etc. Use existing policy input funds to support AI programs to \nmeet conditions, encourage leading and backbone enterprises and industrial innovation \nalliances to take the lead in establishing marketized AI development bases. Use angel \ninvestment, risk investment, start-up investment funds, financial market funding and many \nother such channels to guide social capital to support AI development. Vigorously use \ngovernmental and social capital cooperation and other such models and guide social \ncapital to participate in the implementation of major AI programmes and the \ntransformation and application of scientific and technological achievements.  \n(2) Optimize arrangements to build AI innovation bases \nAccording to the national-level science and technology innovation base arrangements and \nframeworks, comprehensively promote a few internationally advanced innovation bases in \nthe area of AI construction. Guide existing AI-related national focus laboratories, corporate \nnational focus laboratories, national engineering laboratories, and other such bases, and \nconduct research focused on an advanced direction of a new generation of AI. According to \nregulatory procedure, build technological and industrial innovation bases related to the AI \narea with enterprises in the lead, and in cooperation between industry, scholarship, and \nresearch. Give rein to the driving role of leading and backbone enterprises concerning \ntechnological innovation demonstrations. Develop specialized public maker spaces in the \nAI area, stimulate the precise linkage of the newest technological achievements, resources \nand services. Fully give rein to the role of all kinds of innovation bases in concentrating \ntalent, finance, and other such innovation resources; make breakthroughs in basic and \nadvanced AI theory and key common technologies; and launch application \ndemonstrations. \n(3) Comprehensively plan international and domestic innovation \nresources \n  \nSupport domestic AI enterprises to cooperate with international leading AI schools, \nscientific research institutes and teams. Encourage domestic AI enterprises to \"go out,\" \nand provide conveniences and services to powerful AI enterprises conducting foreign \nmergers or acquisitions, share investment, start-up investment, establishing foreign \nresearch centres, etc. Encourage foreign AI enterprises and research institutes to establish \nresearch and development centers in China. With the support of the “One Belt, One Road” \nstrategy, promote the construction of international AI science and technology cooperation \nbases, joint research centres, etc.; accelerate the broad application of AI technologies in \ncountries along the “One Belt, One Road.” Promote the establishment of international AI \norganizations, jointly formulate related international standards. Support related sectoral \nassociations, alliances, and service bodies to build globalized service platforms aimed at AI \nenterprises. \n\n\n25 \nV.  Guarantee Measures \nAiming at the realistic requirements of promoting the healthy and rapid development of AI \nin China, it is necessary to deal with the possible challenges of AI, form an institutional \narrangement to adapt to the development of AI, build an open and inclusive international \nenvironment, and reinforce the social foundation of AI development. \n(1) Develop laws, regulations, and ethical norms that promote the \ndevelopment of AI \n \nStrengthen research on legal, ethical, and social issues related to AI, and establish laws, \nregulations and ethical frameworks to ensure the healthy development of AI. Conduct \nresearch on legal issues such as civil and criminal responsibility confirmation, proteciton of \nprivacy and property, and information security utilization related to AI applications. \nEstablish a traceability and accountability system, and clarify the main body of AI and \nrelated rights, obligations, and responsibilities. Focus on autonomous driving, service \nrobots, and other application subsectors with a comparatively good usage foundation, and \nspeed up the study and development of relevant safety management laws and regulations, \nto lay a legal foundation for the rapid application of new technology. Launch research on AI \nbehavior science and ethics and other issues, establish an ethical and moral multi-level \njudgment structure and human-computer collaboration ethical framework. Develop an \nethical code of conduct and R&D design for AI products, strengthen the assessment of the \npotential hazards and benefits of AI, and build solutions for emergencies in complex AI \nscenarios. China will actively participate in global governance of AI, strengthen the study of \nmajor international common problems such as robot alienation and safety supervision, \ndeepen international cooperation on AI laws and regulations, international rules and so on, \nand jointly cope with global challenges. \n(2) Improve key policies for the support of AI development \n \nImplement tax incentives for small and mid-sized enterprise and startup AI development, \nand, using high-tech enterprises, tax incentives, R&D cost deductions, and other policies, \nsupport the development of AI enterprises. Improve the implementation of open data and \nprotection-related policies, launch open public data reform pilots to support the public and \nenterprises in fully tapping the commercial value of public data, and promote the \napplication of AI innovation. China will study the policy system of education, medical care, \ninsurance, and social assistance to adapt to AI, and effectively deal with the social \nproblems brought by AI. \n(3) Establish an AI technology standards and intellectual property \nsystem \n \nConduct research on strengthening the AI standards framework system.  Adhere to the \nprinciples of security, availability, interoperability, and traceability; and gradually establish \nand improve the basic basis of AI, interoperability, industry applications, network security, \nprivacy protection, and other technical standards. Speed up the promotion of autonomous \ndriving, service robot, and other application sector industry associations in developing \nrelevant standards. Encourage AI enterprises to participate in or lead the development of \n\n\n26 \ninternational standards, and a technical standards \"going out\" approach to promote AI \nproducts and services in overseas applications. Strengthen the protection of intellectual \nproperty in the field of AI, improve the field of AI technology innovation, patent protection, \nand standardization of interactive support mechanisms to promote the innovation of AI \nintellectual property rights. Establish AI public patent pools to promote the use of AI and \nthe spread of new technologies. \n(4) Establish an AI security supervision and evaluation system \n \nStrengthen research and evaluation of the influence of AI on national security and secrecy \nprotection; improve the security protection system of human, technology, material, and \nmanagement support; and construct an early warning mechanism of AI security \nmonitoring. Strengthen the development of AI technology prediction, research and follow-\nup research, adhere to a problem-oriented, accurate grasping of technology and industry \ntrends. Enhance the awareness of risk, pay attention to risk assessment and prevention \nand control, and strengthen prospective prevention and restraint guidance. In the near \nterm focus on the impact on employment, with a long-term focus on the impact on social \nethics, to ensure that the development of AI falls with the sphere of secure and \ncontrollable. Establish and improve an open and transparent AI supervision system, the \nimplementation of design accountability, and application of the supervision of a two-tiered \nregulatory structure, to achieve management of the whole process of AI algorithm design, \nproduct development and results application. Promote AI industry and enterprise self-\ndiscipline, and earnestly strengthen management, increase disciplinary efforts aimed at \nthe abuse of data, violations of personal privacy, and actions contrary to moral ethics. \nStrengthen AI cybersecurity technology research and development, strengthen AI products \nand systems cybersecurity protection. Develop dynamic AI research and development \nevaluation mechanisms, focus on AI design, product and system complexity, risk, \nuncertainty, interpretability, potential economic impact, and other issues. Develop a \nsystematic testing methods and indicators system. Construct a cross-domain AI test \nplatform to promote AI security certification, and assessment of AI products and systems \nkey performance. \n(5) Vigorously strengthen the training of an AI labor force \n \nAccelerate the study of the employment structure brought on by AI, changes in \nemployment methods, and the skills demand of new occupations and jobs, establish a \nlifelong learning and employment training system to meet the needs of the intelligent \neconomy and intelligent society, and support institutions of higher learning, vocational \nschools and socialization training Institutions to carry out AI skills training.  Substantially \nincrease the professional skills of workers to meet the development requirements of \nChina's AI to bring high-quality jobs. Encourage enterprises and organizations to provide AI \nskills training for employees. Strengthen the re-employment training and guidance of \nworkers to ensure the smooth transfer of simple and repetitive workers due to AI. \n(6) Carry out a wide range of AI scientific activities \n  \nSupport the development of a variety of AI scientific activities, encourage the broad masses \nof scientific and technological workers to join the promotion of AI popular science, and \n\n\n27 \ncomprehensively improve the level of the whole society on the application of AI. Implement \na universal intelligence education project. In the primary and secondary schools, set up AI-\nrelated courses, and gradually promote programming education to encourage social forces \nto participate in the promotion and development of educational programming software and \ngames. Construct and improve the AI science infrastructure, give full play to all kinds of AI \ninnovation base platforms and other popular science roles, encourage AI enterprises, and \nresearch institutions to build open source platforms for public open AI research and \ndevelopment, plus production facilities or exhibition halls. Support the development of AI \ncompetitions, encourage the formation of a variety of AI science creational work efforts. \nEncourage scientists to participate in AI science. \nVI.  Organization and Implementation \n \nThe development plan for a new generation of AI is a far-sighted scheme affecting the \noverall picture and the long term. We must strengthen organizational leadership, complete \nmechanisms, take aim at objectives, keep tasks closely in view, realistically grasp \nimplementation with a spirit of hammering nails, and carry out the blueprint to the end. \n(1) Organizational leadership \n \nAccording to the unified deployment of the Party Center and the State Council, the National \nScience and Technology Structural Reform and Innovation System Construction Leading \nSmall Group will take the lead in comprehensive planning and coordination, it will \ndeliberate major tasks, major policies, major questions, and major work arrangements. \nPromote AI-related legal and regulatory construction. Guide, coordinate and supervise \nrelevant departments in carrying out the deployment and implementation of tasks from the \nplan. With the support of the interministerial joint conferences for national science and \ntechnology planning (earmarks, funding, etc.) management, the Ministry of Science and \nTechnology will, together with relevant departments, be responsible for moving forward the \nimplementation of major science and technology programmes for a new generation of AI, \nand strengthen linkages and coordination with other programmatic tasks. Establish an AI \nPlan Implementation Office. This office will be part of the Ministry of Science and \nTechnology and will be concretely responsible for moving the implementation of the plan \nforward. Establish an AI Strategy Advisory Committee, to research major far-sighted and \nstrategic questions concerning AI and to provide advice and assessment concerning major \npolicy decisions on AI. Move forward with the construction of an AI think tank, support all \nkinds of think tanks to launch research on major AI questions, and provide strong and \npowerful support for the development of AI. \n(2) Guarantee implementation \n \nStrengthen the deconstruction of plan tasks, clarify responsible work units, schedules and \narrangements, formulate annual and phase-type implementation plans. Establish \nmonitoring and evaluation mechanisms for the implementation situation of the plan, such \nas annual assessment and intermediate evaluation. Adapt to the characteristics of the \nrapid development of AI, and strengthen dynamic adjustment of plans and programs on the \n\n\n28 \nbasis of the progress of tasks, the completion of intermediate objectives, new trends in \ntechnological development, etc. \n(3) Trials and demonstrations \n \nWe must formulate concrete plans for major AI tasks and focus policy measures, and \nlaunch trials and demonstrations. Strengthen comprehensive guidance over trials and \ndemonstrations in all departments and all localities, quickly summarize and disseminate \nreplicable experiences and methods. Advance the healthy and orderly development of AI \nthrough advance trials and guiding demonstrations. \n(4) Public opinion guidance \n \nFully use all kinds of traditional media and new media to quickly propagate new progress \nand new achievements in AI, to let the healthy development of AI become a consensus in \nall of society, and muster the vigor of all of society to participate in and support the \ndevelopment of AI. Conduct timely public opinion guidance, and respond even better to \nsocial, theoretical, and legal challenges that may be brought about by the development of \nAI. \n  \n###","difficulty":"hard","domain":"Multi-Document QA","length":"short","question":"Which of the following statement is false according to the three materials?","sub_domain":"Governmental"}

Source: https://huggingface.co/datasets/zai-org/LongBench-v2

initial import

Posting: /agents

GET /api/v1/write?intent=publish&task_id=b6b90ce0-0d61-5ebe-8a42-cef2fbed4f4f&body={url_encoded_text}&agent_name={optional_name}&nonce={optional_random_id}