Google hired back the man Thinking Machines fired, for post-training
Barret Zoph returns as a VP of research after a year that ran Google to OpenAI to Thinking Machines to OpenAI, and Google's statement names exactly what it is buying.
Google said on 26 August that Barret Zoph is joining as a vice president of research, returning to the company where he was a research scientist from 2016 to 2022. In between he co-founded Thinking Machines Lab, was fired from it in January, went back to OpenAI, and left again five months later. Erin Woo, who broke the story for the Wall Street Journal, called it the completed boomerang.
this just in: Barret Zoph, who dramatically left Thinking Machines Lab to go to OpenAI, is now going to Google as a VP of research on RL/posttraining
completing the boomerang of Google --> OpenAI --> TML --> OpenAI --> Google
scoop with @MeghanBobrowsky t.co/1tASQhViGj
Google's statement names the job precisely. "We look forward to Barret returning to Google and bringing his RL and post-training expertise to Gemini," a spokesman said. Not research generally. Reinforcement learning and post-training, on one product (Gemini is named; DeepMind is not).
That is the whole story, and it is a procurement story more than a personnel one.
So consider what Google is buying, and what it is accepting to get it. Mira Murati fired Zoph in January and told staff he had shown a pattern of performance, conduct and trust problems, the Journal reported; while at Thinking Machines he had an undisclosed relationship with a colleague, and his responsibilities were scaled back. Zoph's account differs. He says Murati dismissed him only after he had already told her he was leaving for another job, and that the reduction in his duties was temporary. Both versions are on the record and we have no way to adjudicate between them (we would not trust anyone who claims otherwise).
But what is not in dispute is that a company with Alphabet's legal department read all of it and hired him anyway.
Price that. Earlier this year Noam Shazeer left for OpenAI and John Jumper, a Nobel laureate, left for Anthropic, and Alphabet stock fell as much as 7% in a single session. Jeff Dean left in early August after 27 years to start a company with three senior colleagues. Demis Hassabis moved from DeepMind chief executive to chairman, with Koray Kavukcuoglu promoted to run Gemini development. A 7% move on two departures is the market saying, out loud, that a handful of named individuals are material information about one of the largest companies on earth.
Our read: this hire tells you where Google thinks the Gemini gap is. Probably not compute, which it has more of than almost anyone. Not pretraining, where it has been strong for a decade. Post-training and reinforcement learning, the layer that turns a capable base into a product people prefer, and the layer where OpenAI and Anthropic have kept an edge. And you probably do not buy a VP of research with a public history of trouble unless the alternative looks worse.
The counter is that we are over-reading one line of corporate boilerplate. Spokespeople name the hot area because it reassures investors, hires happen for a dozen reasons, and Zoph is good at this work regardless of what any statement says. Fair enough. So here is the test: if Google's next Gemini release is sold on post-training and RL gains rather than on context, price or scale, the boilerplate was accurate. If it leads on compute, we misread it.
Spare a thought for Thinking Machines Lab, eighteen months old and now missing two of its founders. Luke Metz left OpenAI too, for Meta. When the whole industry can outbid you for the people who started your company, what exactly did you found?