John Schulman

there's a particular kind of person in ai right now who seems constitutionally unable to stay still — not out of restlessness exactly, but because the field keeps outrunning the institutions built to contain it. John Schulman is that person.

he grew up in the late eighties, came up through academic machine learning, and landed at a moment when the field was transitioning from a research curiosity into something with genuine cultural weight. he was one of the co-founders of OpenAI back when that still meant something scrappy and uncertain, and his specific contribution — reinforcement learning from human feedback, the technique that sits underneath basically every modern conversational AI — is the kind of work that doesn't announce itself loudly but quietly restructures what's possible. RLHF is the reason chatgpt feels like it's talking *to* you rather than *at* you. that's his fingerprint.

but what's making him a name to follow right now isn't what he built at OpenAI. it's what he's done since leaving.

in August of 2024, he announced he was joining Anthropic — OpenAI's closest rival, founded by people who had also left OpenAI, which tells you something about the centrifugal forces inside that organization. the move felt significant. Anthropic is where a certain kind of safety-minded researcher goes when they want to feel like the work still has a moral architecture. Schulman seemed like a fit. and then, less than six months later, he was gone again.

in February 2025, he surfaced at something called Thinking Machines Lab, taking the chief scientist role. the company was founded by Mira Murati — herself a former OpenAI executive, formerly its CTO — and it's offering the kind of compensation packages, reportedly up to half a million dollars in salary, that signal serious institutional ambition rather than a side project. Schulman wasn't the only prominent name drawn there; the lab has been assembling a team with real pedigree, even as the broader AI talent market looks increasingly like a game of musical chairs. Barret Zoph, who had been CTO at Thinking Machines, has since returned to OpenAI — a reminder that even within this orbit, loyalty is provisional and trajectories are non-linear.

what makes Schulman interesting to watch isn't the job-hopping as spectacle. it's what the movement suggests about where the real intellectual action might be. OpenAI is now a very large, very scrutinized organization with enormous commercial pressure. Anthropic has carved out a particular lane — thoughtful, cautious, philosophically careful — but it's still operating within the logic of the established players. Thinking Machines Lab is something different: a bet that the people who shaped the first wave of this technology have something new to build that doesn't fit inside the structures they came from. Schulman showing up there as chief scientist is a signal about what kind of work he thinks is worth doing next.

he's someone who has been, at nearly every juncture, at the technical center of something that mattered — and who keeps choosing the next thing over the comfortable thing. the index doesn't care much about credentials. it cares about where attention is actually moving, and right now, attention is moving toward what Schulman is going to build next.