The Oxford Martin Working Paper Series on Technological and Economic Change

Carl Benedikt Frey, Giorgio Presidente and Pia Andres. Data-Biased Innovation: Directed Technological Change and the Future of Artificial Intelligence. The Oxford Martin Working Paper Series on Technological and Economic Change

This paper examines how privacy regulation has shaped the trajectory of artificial intelligence (AI) innovation across jurisdictions. We construct a novel taxonomy of AI technologies based on their data intensity and analyze patent applications from 57 countries across 76 industries from 2010 to 2021. Our descriptive analysis reveals three key patterns: a substantial shift from data-saving to data-intensive AI methods over the 2010s, increasing market concentration in innovation among established firms, and pronounced geographic heterogeneity in both innovation output and technological focus around the world. Exploiting variation in firms’ exposure to the European Union’s General Data Protection Regulation (GDPR), we find that exposed applicants significantly altered their technological trajectories. Applicants with greater exposure to EU markets increased their data-saving patents while decreasing data-intensive ones relative to the pre-GDPR period. This effect is driven primarily by EU-based firms. Additionally, the GDPR appears to have reduced overall AI patenting in the EU while reinforcing the market dominance of established companies.