Alex Imas is a behavioral economist and Professor of Behavioral Science and Economics at the University of Chicago Booth School of Business. His recent work extends classical behavioral-economics methods to the principal-(AI)agent setting.
The DeepMind Institute's September 2026 biography gives different titles: Director of AGI Economics at Google DeepMind and Professor of Economics and Applied AI at the University of Chicago. Under those titles he co-authored, with Julian Jacobs, "Economic policy for AGI", one of the institute's inaugural essays Economic policy for AGI (Jacobs and Imas, DeepMind Institute, 2026). See DeepMind Institute.
Background and methods
Imas's research applies behavioral-economics tools to AI-mediated interaction, including induced-values protocols, social-preference elicitation, and Big Five trait analysis. His prior work on social preferences, gender differences in negotiation, and prompt-elicitation methodology forms the methodological basis of his 2025 study on AI-mediated bargaining.
Agentic Interactions study
Imas is lead author of "Agentic Interactions" (Imas, Lee, Misra, December 2025), an experimental marketplace study co-authored with Kevin Lee and Sanjog Misra on the persistence and amplification of human heterogeneity in AI-mediated markets. The paper reports that 73% of variance in AI-mediated bargaining outcomes loads on individual fixed effects of the human principal, and documents 16.5% higher outcome variance under AI mediation than under human-to-human bargaining. It finds that AI mediation reverses the gender gap in negotiation: female-prompted Seller agents do better than male-prompted ones, despite the agents not seeing principal demographics. The study introduces machine fluency as a source of inequality and is a companion to Misra (2025)'s theoretical framework of specification hazard.
The study connects to questions of AI-mediated bargaining outcomes, algorithmic fairness in agent-mediated markets, the distributional effects of AI delegation, and the empirical study of the principal-(AI)agent problem.
Relationships
- related: Machine Fluency — co-introduces the concept
- related: Specification Hazard — co-introduces the concept
- related: Principal-Agent Problem Applied to AI
- related: DeepMind Institute — inaugural essay author
- related: Agentic AI
- related: AI Divides (Literacy / Occupational / Ethico-Philosophical) — machine fluency adds a new dimension
- supports: Agentic Interactions — Imas, Lee, Misra (2025) — lead author
- related: Sanjog Misra — co-author
- related: John J. Horton — adjacent agent-economy research cluster