Major plot twist at OpenAI, co-founder exits

CHENNAI: A day after OpenAI announced major updates to its ChatGPT chatbot, the company said its chief scientist and co-founder was leaving. There were signs that Ilya Sutskever would quit, six months after he helped lead the rebellion that briefly ousted Sam Altman as OpenAI’s C.E.O. (He hasn’t been spotted in the office since that episode.) But it also raises questions about the future of a leading developer of generative A.I.

“OpenAI would not exist without him and certainly was shaped by him,” Altman told The Times about Sutskever. It’s hard to understate Sutskever’s importance to OpenAI: He helped found it in 2015 along with Altman and others including Elon Musk, and his stature as a leading researcher in neural networks gave the fledgling company instant credibility.

But Sutskever’s presence at OpenAI may have become untenable. In November, when he was a board member, Sutskever teamed up with other directors to fire Altman, accusing him of not being “consistently candid in his communications.” After a slew of OpenAI employees resigned in protest, he had a change of heart, effectively stepping down from the board and supporting

Altman’s return. After Altman’s reinstatement, Sutskever has stayed quiet publicly though he did spearhead the creation of a so-called Super Alignment team to help ensure that OpenAI’s products didn’t harm humanity.

(Jan Leike, who ran that team with Sutskever, also resigned on Tuesday and will be replaced by another company co-founder, John Schulman.) Meanwhile,

OpenAI had already effectively elevated Jakub Pachocki as chief scientist. Sutskever’s exit is another sign that Altman is in the driver’s seat. While OpenAI is expanding its board after last year’s turmoil, Altman remains the company’s most prominent figure. (Pachocki and Schulman are regarded as his allies.) While Sutskever has worried about A.I.’s apocalyptic potential, his statement on Tuesday said he was confident that the company would build artificial general intelligence A.I. as sophisticated as the human brain “that is both safe and beneficial.” That’s in some ways a validation of Altman’s approach of speedy innovations and commercialization.