We analyze the sectoral and national systems of firms and institutions that collectively engage in Artificial Intelligence (AI). Moving beyond the analysis of AI as a General Purpose Technology, or its particular areas of application, we draw on the evolutionary analysis of sectoral systems, and ask “Who does what” in AI. We provide a granular view of the complex interdependency patterns that connect developers, manufacturers, and users of AI. We distinguish between AI enablement, AI production, and AI consumption and analyze the emerging patterns of co-specialization between firms and communities. We find that AI provision is characterized by the dominance of a small number of Big Tech firms, whose downstream use of AI (e.g. search, payments, social media) has underpinned much of the recent progress in AI, and who also provide the necessary upstream computing-power provision (Cloud and Edge). These firms dominate top academic institutions in AI research, further strengthening their position. We find that AI is adopted by, and benefits, the small percentage of firms that can both digitize and access high-quality data. We consider how the AI sector has evolved differently in the three key geographies—China, the US and the EU—and note that a handful of firms are building global AI ecosystems. Our contribution is to showcases the evolution of evolutionary thinking with AI as a case study: We show the shift from national/sectoral systems to triple-helix/innovation ecosystems and digital platforms. We conclude with the implications of such a broad evolutionary account for theory and practice.
Read the full paper here.