Innovation and Artificial Intelligence: A Comparative Perspective

• In Event: Artificial Intelligence and Challenges in Policymaking

Abstract

This paper uses a comparative lens to examine innovation policy in the Republic of Korea, Sweden and Canada with respect to Artificial Intelligence. Specifically, the research question to be pursued is how have public policies regarding Artificial Intelligence in Korea, Sweden and Canada impacted innovation development in this domain?

While from a political economy perspective all three countries are small, export-oriented nations, the Republic of Korea is a recognized innovation leader (OECD 2023) as is Sweden (OECD 2016). However, Canada has had much less success on this front and indeed its innovation polices have tended to be fractured and less targeted over time (Scharf 2022 unpublished). This case study is intended to determine if these more general trends prevail regarding innovation initiatives with respect to AI.

Theoretical Framework
The theoretical framework for the case study is the Developmental Network State (DNS), as developed by Block (2008), Keller (Block and Keller 2011; Block, Keller and Negoita 2020, 2023), Mazzucato (2015, 2018, 2023) and Ó Riain (2004, 2014). The framework focuses on three key pillars of activist public policy – mission, targeted resourcing and thickening of triadic networks among business, academia and government. The DNS literature, from a comparative perspective, is still relatively underdeveloped. While it has been used to study Ireland, Chile, China, Canada and, to a certain extent Israel and Taiwan, this would be the first effort to use the framework to examine AI as a case study for Korea, Sweden, and Canada.
The paper will argue that the choice of DNS stands as an important alternative to Varieties of Capitalism (VOC) literature in bringing a comparative lens to this issue, avoiding the more binary and static nature of VOC in dealing with policy over time. It will also briefly discuss how the DNS is distinct from developmental state literature.

Methodology
In terms of methodology the time period for this study is an historical perspective, spanning the years from 1987 to 2023. The research employs qualitative sources as well as critical economic data. Policy priority-setting is accessed through legislation, key strategic plans and roadmaps, annual reports of key innovation agencies, as well as expert studies initiated by government regarding AI. Most importantly, however, the study is crafted to undertake a “quasi-experimental” approach (Campbell 1969; Cook and Campbell 1986). Specifically, the research evaluates if governmental priority-setting and policy implementation -- through policy durability and mission over time, enhanced ecosystems and resourcing -- affect rates of innovation.
Based on this technique as developed by Cook and Campbell, the article argues that in tracking policy initiatives over an “extended times series” this can be then related to policy impacts – in this case, rates of innovation. Consequently, where applicable in the country comparisons, this research has gone back in history to assess when the first public policy interventions on AI occurred. Further, it examines whether late-starter nations have been able to narrow the gap on AI and what are the implications for governance and innovation.

Three Key Pillars of the Case Study
The article then focuses on the three key dimensions or pillars of the case study, as laid out in the theoretical framework. Firstly, it traces out the policy durability and mission of AI innovation policy in the respective countries over time, analysing whether this has been an ongoing priority. Secondly, it examines the extent to which resourcing has been targeted to these initiatives around AI and whether that has been sustained. Thirdly, it examines whether policy has focused on thickening robust innovation ecosystems for AI – and whether policy intention has actually been channelled into realistic implementation in the AI field.

Innovation Impacts
Finally, the paper turns to the actual economic indicators – AI patent data, estimated Venture Capital investments in AI startups – that can be tracked and what this suggests for the causal implications of respective national innovation policies.

Conclusion
The paper concludes in assessing the comparative differences that have emerged in AI public policy initiatives among the respective countries. It also evaluates whether this aligns with the usual assumptions around Korea and Sweden as innovation leaders and Canada as more of a laggard. It also reflects on the value that DNS may bring to comparative analysis in this field.

Author

• Shirley Anne Scharf, Centre for International Governance Innovation