OpenAI co-founder resigns after 10 long years of being with the team and is set to be replaced by Jakub Pachocki.
Ilya Sutskever, a co-founder and longstanding chief scientist at OpenAI, has departed from the organization.
Tuesday evening, OpenAI CEO Sam Altman announced the news in a post on X.
Altman stated, “This saddens me greatly; Ilya is undoubtedly one of the brightest minds of our generation, an authority in our field, and a dear friend.” With him, OpenAI is the organization that it is today. “Although he is leaving to work on something more personally significant, I will be eternally grateful to him for his efforts here and his dedication to completing the mission we began together.”
OpenAI’s director of research, Jakub Pachocki, has taken over for Sutskever. Pachocki began working for OpenAI in 2017 as a research lead on the Dota team, which developed an AI system capable of defeating human participants in the Dota 2 strategy game developed by Valve. Principal of Research Pachocki was promoted from research director at OpenAI’s reasoning and science of deep learning organizations to research principal.
It initially needed to be clarified whether Pachocki would also assume leadership of OpenAI’s Superalignment team, an organization that Sutskever and Jan Leike had been in charge of until now. According to The New York Times, Leike is also resigning from OpenAI.
In July, OpenAI established the Superalignment team to devise methods to guide, oversee, and administer “superintelligent” AI systems—hypothetical entities possessing intelligence that surpass human beings. According to The Times, John Schulman, an additional co-founder of OpenAI, will assume the position of overseer.
According to TechCrunch, the Superalignment team will be “deeper integrated” throughout OpenAI’s research to “improve the achievement of its objectives.” That means the team’s current configuration may transform in the coming years.
OpenAI’s president, Greg Brockman, wrote on X that Sutskever “significantly contributed to laying the groundwork for what OpenAI is today.”
Sutskever’s departure, which follows the introduction of OpenAI’s most recent premier generative AI model, GPT-4o, and significant enhancements to the company’s widely-used AI-powered chatbot ChatGPT, effectively concludes an eventful sequence that commenced in November of last year.
Sutskever and OpenAI CTO Mira Murati confronted members of OpenAI’s previous board of directors approximately one week before Thanksgiving to convey their apprehensions regarding Altman’s conduct. Disagreements regarding OpenAI’s trajectory were purportedly at issue; Sutskever allegedly became impatient with Altman’s haste to introduce AI-powered products while neglecting safety efforts.
The previous board, comprised of Sutskever, decided to terminate Altman immediately without informing nearly anyone, including most of OpenAI’s staff. According to a statement from the board, Altman had not been “consistently candid” in his communications with board members.
Microsoft and other OpenAI investors were enraged by the decision, which jeopardized the company’s stock sale and prompted a remarkable reversal: most OpenAI employees, including Sutskever, vowed to resign unless Altman was reinstated immediately.
In the end, Altman was reinstated, and a significant portion of the previous board members tendered their resignations. The Times reports that Sutskever never returned to work after that; since November, Pachocki has served effectively as chief scientist.
Sutskever, who graduated from the University of Toronto with a doctorate in computer science and labored under the guidance of AI luminary Geoffrey Hinton, joined OpenAI in 2015, following his departure from Google Brain, an AI research division of Google. Sutskever has made significant contributions to artificial intelligence, including ImageNet, one of the earliest modern computer vision systems, and AlphaGo, a game-playing AI system developed by DeepMind.
After that, what will he do? Sutskever is not prepared to reveal. However, he stated in a statement on X that he is departing OpenAI with the conviction that the organization will develop “beneficial and secure” artificial general intelligence—AI capable of performing any task a human can.
Sutskever stated, “I am looking forward to the next project, which is extremely personal and will provide additional information in due course.” “Working together [at OpenAI] was a tremendous honor and privilege; I sorrowfully say farewell to all involved.”
