A Fields Medal-winning mathematician who fears AI—and went to work at OpenAI

Jacob Zimmerman is one of the greatest mathematicians of his generation and a Fields Medalist. And he openly admits that he fears artificial intelligence. A logical question arises: Why, then, would he choose to work at OpenAI?

This isn’t cognitive dissonance—it’s a calculated move. If you believe that the technology poses a systemic risk, sitting on the sidelines and issuing warnings is a weak strategy. The influence over the trajectory of development shifts to where decisions are made about scaling, standardization, and access to models.

It’s a familiar pattern. Security researchers who publicly criticized the pace of deployment have moved into the labs—not because they changed their minds, but because the alternative is even worse. When the threat is technical, we need not only policymakers and regulators, but also people capable of understanding the mathematics of alignment at a level comparable to that of the models themselves.

Zimmerman is a top-tier mathematician. If he sees a structural problem in AI that can be formalized, his presence at OpenAI is more valuable than yet another article full of warnings. The question is this: how many such people are needed internally to actually change practices, and will they have veto power—or only an advisory role?

Source: Hacker News - Newest: ""AI" "LLM""