Technology/Market Forces in Developing General Purpose Technology across Leading and Fast-follower Countries: The Case of AI

• In Event: AI Governance and Regulation: From State to Federal

Abstract

General Purpose Technologies (GPTs), such as Artificial Intelligence (AI), represent transformative technological architectures that reshape economies, industries, and scientific fields. While the innovation literature emphasizes the dual roles of technology-push and demand-pull forces in driving technological change, less is known about how these drivers influence GPT development within the global academic research landscape, particularly how researchers in different national contexts differentially respond to technological disruption. This gap is consequential: understanding the mechanisms by which academic knowledge production aligns with technological frontiers has direct implications for how nations allocate research resources and cultivate next-generation scientific talent. This study investigates how technological and demand-side forces induce academic researchers, specifically graduate students, who possess high cognitive flexibility and low path-dependency to engage with AI research. Utilizing a novel dataset of computer science dissertations from U.S. and South Korean universities, we examine dissertation topic selection following two pivotal events: the 2017 introduction of the Transformer architecture (technology-push) and the 2022 launch of ChatGPT (demand-pull). We hypothesize that technology-push forces primarily drive GPT development in leading countries (e.g., the U.S.), whereas demand-pull forces are the primary drivers in fast-follower countries (e.g., Korea). Our findings suggest that abundant research resources and research-intensive institutes amplify the technology-push effect in leading countries. Conversely, limited resources and non-research-intensive institutes strengthen the demand-pull effect in fast-follower contexts. Theoretically, this study advances the innovation literature by demonstrating that the relative salience of technology-push versus demand-pull forces is not uniform across national contexts, but is systematically conditioned by a country's position in the global academic research hierarchy. This reframes the push-pull dichotomy as a structurally embedded phenomenon, rather than a universal feature of GPT diffusion. These findings carry meaningful implications for education and R&D policy. For leading countries, sustaining institutional capacity at research-intensive universities is critical to maintaining frontier-pushing innovation. For fast-follower countries, demand-pull dynamics, while often undervalued in policy discourse, may represent a strategically viable pathway for rapid AI capability development, particularly when frontier research resources are constrained. Policymakers designing graduate education programs, research funding schemes, and national AI strategies should account for these asymmetric mechanisms, rather than applying uniform frameworks across structurally distinct national contexts.

Author

• Eunji Emily Kim, University of California San Diego
  Presenting Author