Google DeepMind has hired University of Chicago behavioral economist Alex Imas as Director of AGI Economics, a new role announced in early May via X. Imas will report directly to Shane Legg, the lab's chief AGI scientist and co-founder, and lead a team building economic models for a world where artificial general intelligence is no longer hypothetical.
The hire
Imas, on paper, is an interesting pick. He's the Roger L. and Rachel M. Goetz Professor at Chicago Booth, where his work sits at the intersection of behavioral economics and applied AI. He recently co-authored The Winner's Curse, an update of Richard Thaler's 1992 anomalies book, with Thaler himself. Not a tech-industry economist parachuted in to make announcements look serious. He has actual chops in how humans make messy, irrational decisions, which is presumably relevant if you're trying to model an economy where software agents are doing more of the deciding.
Not the first lab to try this
DeepMind isn't out front here. OpenAI hired Duke's Ronnie Chatterji as chief economist in 2024, and last year added Harvard's Jason Furman, a former Obama advisor, to research AI and jobs. Anthropic has convened a panel of ten economists doing similar work. That all three frontier labs have separately decided they need a heavy bench of economic thinkers should tell you something, though what exactly is open to interpretation.
Daron Acemoglu, the Nobel laureate who has been studying AI's labor impact since 2018, gave MIT Technology Review his read on the trend. "What I hope we won't get," he said, "is that they're interested in economists just to further their viewpoints or further the hype." Which is the polite academic way of saying these hires could be either real research or expensive lobbying.
What he'll actually work on
The team's stated agenda covers labor market shifts, capital redistribution, institutional adaptation, and the behavior of autonomous AI agents in markets. The job listing, still posted on DeepMind's careers page, asks the hire to question existing assumptions about scarcity, wealth, and distribution, and to build agent-based simulations of post-AGI scenarios. Whether you find that ambitious or premature depends on how much of the AGI timeline you've already bought into.
Imas's recent academic work tracks the last item closely. A working paper on his research page examines how AI agents inherit and even amplify the biases of the humans who deploy them, with sorting based on principal characteristics like gender and personality. The agent economy, in his framing, is not the clean rational-actor model of classical economics. Which is roughly behavioral economics, applied to a new substrate.
How much to read into the "post-AGI" framing is the other question. Treating AGI's arrival as given when you're designing a research agenda is itself an assumption worth flagging, especially from a lab whose own leadership has been forecasting it for years.
What's next
DeepMind's job listing for the team is still live, meaning more hires are coming. Initial research output from the group is likely months out.
