Search
AI & Society Animated Illustration

Building an AI World:
Report on National and Regional AI Strategies
Second Edition

Introduction

With Russia’s publication of its National Strategy for the Development of Artificial Intelligence (AI) in October 2019, and Brazil’s announcement in December 2019 that the country was opening public consultations for its national AI strategy, all 15 of the world’s largest economies have now developed, or are in the process of developing, strategies to guide their governments’ policy approach towards the use and development of AI.

In November 2018, CIFAR published the first edition of Building an AI World: Report on National and Regional AI Strategies. Since that time, a number of countries have further developed or finalized their own national AI strategies, and more have taken steps towards creating such strategies to ensure that they, too, can capitalize on AI to drive growth and remain competitive. This second edition provides an updated overview of the global landscape of AI strategies as of January 2020.

This report aims to provide a snapshot of the rapidly evolving area of technological policy, highlighting the variety of approaches taken by different countries, and providing policymakers with a reference and a toolkit of potential measures to shape the development of the field in their own countries. It does not seek to evaluate the various national strategies or attempt to draw causal links between the strategies and the current state of development of AI in the respective country.

 

Lead author: Johnny Kung, PhD

Additional research by:
Gaga Boskovic and Charlotte Stix

 

 

Key Findings

  • By the end of January 2020, 28 national or supranational jurisdictions — including 27 countries as well as the European Union (EU) — have published coordinated strategies to guide their governments’ policies on AI. An additional 18 countries are developing their own national AI strategies.
  • Countries that are developing or have developed national AI strategies continue to be concentrated among the advanced economies of Western Europe, North America and East Asia, but there is a notable increase in the number of emerging or developing economies on this list.
  • Among the AI strategies published between November 2018 and January 2020, the majority either do not come with new or additional funding, or only include new funding for a few specific policies.
  • Compared to strategies described in the first edition of this report, the new strategies tend to be more comprehensive, incorporating policy measures that address most or all of eight major policy areas (research, talent development, skills, industrial policy, ethics, data & digital infrastructure, AI in government, and inclusion).
  • The published AI strategies tend to have the most specific policy measures to address data & digital infrastructure, talent development and industrial policy, and the least specific measures for AI in government and inclusion.
  • Broadly speaking, the published national AI strategies can be grouped into three main categories: those that are largely focused on AI research and development, those that are comprehensive across policy areas but with less specific policy measures, and those that are comprehensive and specific.

 

National AI strategies in 2019

An Increasingly Globalized Landscape

Since the publication of the first report, 16 jurisdictions (15 countries plus the EU) have published national strategies on AI, for a total of 28 as of January 2020. An additional 18 countries have started, or have made progress on, the development of a national strategy. While the majority of nations with or working on AI strategies continue to be concentrated in the advanced economies of Western Europe, North America and East Asia, what is notable is that a number of emerging or developing economies in the rest of Asia, Eastern Europe and Latin America (such as Brazil, Qatar, Russia and Sri Lanka) have joined this club in the past year.

Similar to strategies reported last year, the strategies in this report can be broadly divided into two groups: those with specific funding when first announced and those without. In fact, the majority of new strategies did not have new or additional funding beyond what the respective governments already invest in, e.g., research grants, or only included new funding for a few specific policies in the strategy.

 

Figure 1
Current landscape of AI strategies

This world map provides an overview of the countries, as of January 2020, that have either published a national AI strategy, with or without full funding specified upon release, or have a strategy in development.

dots-red  Funding specified upon strategy’s release

dots-blue  Funding (new/extra) not specified, or only for specific measures

dots-yellow  Strategy in development

 

A Trend Towards More Comprehensive Strategies

This report analyzes the various national strategies according to eight broad policy areas. Common or notable measures across the published AI strategies to address each of the policy areas include:

A Trend Towards More Comprehensive Strategies
Policy Area Common or Notable Policy Measures

Scientific Research

  • Establish national AI research centres
  • Increase investment in AI research

AI Talent Development

  • Remuneration incentives and visa policies to attract international talent
  • Increase AI programs or components in master’s and PhD programs

Skills and the Future of Work

  • Increase reskilling / training programs for workers
  • Incorporate more STEM (including AI) in primary-to-undergrad curriculum

Industrialization of AI technologies

  • Establish digital innovation hubs to connect companies to AI expertise
  • Use state investment funds to support startups and leverage private investments

Ethical AI Standards

  • Establish guidelines and promote research on explainability and accountability

Data and Digital Infrastructure

  • Make public datasets available for development of AI tools
  • Set up regulatory sandboxes to test AI products
  • Develop tools in local languages

AI in the Government

  • Pilot AI-based solutions in public service

Inclusion and Social Well-Being

  • Support designs and tools that reduce bias and discrimination

 

In evaluating the emphasis placed by each national strategy on the eight policy areas, the current report produces a “specificity” value for each policy area, which enables a semi-quantitative evaluation of the comprehensiveness of each strategy — e.g., whether it proposes detailed measures in multiple policy areas, or if it focuses only on a certain number of policy areas (Figure 2). Across all 28 published national strategies, the policy areas with the most specific measures are data and digital infrastructure, talent development, and industrial policy, while those with the fewest specific measures are AI in government and inclusion (Figure 3).

 

Figure 2
Radar plot for AI strategies of four example jurisdictions, showing the specificity value (0-5, with 5 being the most specific) across 8 policy areas.

Canada

 

This radar plot for Canada’s national AI strategy shows specificity values of 5 for Research, 5 for Talent, 0 for Future of Work, 1 for Industrial Policy, 3 for Ethics, 0 for Data and Digital Infrastructure, 0 for Government, and 0 for Inclusion.

 

Germany

 

This radar plot for Germany’s national AI strategy shows specificity values of 5 for Research, 5 for Talent, 5 for Future of Work, 5 for Industrial Policy, 5 for Ethics, 4 for Data and Digital Infrastructure, 1 for Government, and 5 for Inclusion.

 

European Union

 

This radar plot for the European Union’s AI strategy shows specificity values of 3 for Research, 4 for Talent, 3 for Future of Work, 5 for Industrial Policy, 5 for Ethics, 5 for Data and Digital Infrastructure, 5 for Government, and 4 for Inclusion.

 

Qatar

 

This radar plot for Qatar’s national AI strategy shows specificity values of 2 for Research, 3 for Talent, 5 for Future of Work, 4 for Industrial Policy, 4 for Ethics, 5 for Data and Digital Infrastructure, 1 for Government, and 0 for Inclusion.

 

Figure 3
Proportion of published national strategies with a specificity value of 3 or above in each policy area (counting only those in which the policy area is mentioned).

This bar graph shows the policy areas, in order from the highest to the lowest proportion with specific measures, are Data and Digital Infrastructure, AI Talent, Industrial Policy, Ethics, Research, Future of Work, AI in Government, and Inclusion.

 

To assess similarities across national strategies in terms of their areas of focus or comprehensiveness, an unsupervised hierarchical clustering was performed (Figure 4). The published strategies fall into three major clusters: ones that are largely focused on AI R&D (research, talent development and industrial policy), those that are comprehensive across policy areas but with less specific policy measures, and those that are comprehensive and specific. Generally speaking, many of the strategies that were released before the first edition of this report (including those of Canada, Japan, Singapore, South Korea and Taiwan) fall into the first cluster, whereas the new national strategies surveyed in this report tend to be more comprehensive, outlining measures in most or all of the eight policy areas.

 

Figure 4
Heatmap of the specificity values across policy areas for all AI strategies. Strategies are ordered by hierarchical clustering

Roughly speaking, one cluster of strategies are comprehensive and have specific policies, the second cluster of strategies are comprehensive but less specific, and the third cluster of strategies are focused on AI research and development.

 

Conclusion

As advances continue to be made in both basic AI research and the development of technology that applies AI to a variety of sectors, countries around the world are increasingly recognizing the importance of creating comprehensive, coordinated national strategies in order to remain competitive with their peers. More and more of these strategies not only advance AI research and development, but also seek to prepare the broader society — both the public and private sectors as well as the wider population — for the economic, social, ethical and policy implications of AI. In the coming year, even more countries, from a wider diversity of geography and economic development, are anticipated to finalize their own national AI strategies. Continued tracking and analysis of these policies will be an informative and critical exercise to evaluate the global approaches to this emerging field of technology and policy, and to identify areas where international conversations or coordinated efforts may be warranted.

With more countries adopting national AI strategies, and some of these strategies entering their third or fourth year of implementation, a number of interesting policy questions may benefit from further research and analysis. For example:

  • How much have governments actually followed through on the published strategies and aligned resources or developed policies to support these strategies?
  • What are the accountability mechanisms being put in place to track the implementation of the strategies and evaluate the value in creating such strategies?
  • What kind of impact has the adoption of such strategies made on talent recruitment, research output or economic activities in the respective countries?
  • Does the inclusion or exclusion of certain policy areas make a difference in terms of the strategy’s impact or outcomes?

Tackling these and similar questions will be important for shaping future policy approaches to AI and other areas of rapid technological advancement.