The AI Divide: Evidence from Transaction Data
COM3 Level 2
SR21, COM3 02-60

Abstract:
Whether generative AI narrows or widens economic inequality depends first on who adopts it. We ask whether there is an AI divide across U.S. communities and, if so, whether it reflects what households can afford or what they can do with the technology. Using anonymized consumer card transaction data from 2023 to 2025, we identify recurring paid subscriptions to three providers, OpenAI, Anthropic, and Perplexity, covering 816,000 accounts and $131 million in spending, and link them to census-tract characteristics from the American Community Survey. We find that paid adoption has grown rapidly but is far from widespread, with about three-quarters of tracts recording no paid subscription in our data in 2025, and is concentrated in higher-income, more educated tracts with greater AI-relevant employment. The income elasticity of paid activity is about 0.7 when income is considered alone, but it falls to near zero once the college-educated share and the share of workers in AI-relevant occupations are included, and both of these remain strongly positive. Conditional on these characteristics, tracts with larger Black, Hispanic, and Asian population shares show more paid activity, not less. The evidence points to an AI divide that follows human capital and the task content of local work more closely than purchasing power. Even when a technology is widely accessible at low or no cost, its adoption can remain unequal because the skills and opportunities to benefit from it are unequally distributed.
Bio:
Professor Il-Horn Hann teaches Managing Digital Business Markets at the Robert H. Smith School of Business at the University of Maryland. His research interest is in the intersection of digital innovations and markets. Il-Horn's papers have been published in Management Science, Journal of MIS, and Communication of the ACM. He served on the editorial board of Information Systems Research and on the editorial board of Management Science. Il-Horn received his MA and PhD from the Wharton School of the University of Pennsylvania. He is the co-director of the school's Center for Digital Innovation, Technology and Strategy (DIGITS).

