Following the departure of the group’s two leaders, including OpenAI co-founder and chief scientist Ilya Sutskever, the company has effectively dissolved a high-profile safety team tasked with guaranteeing the security of potential future ultra-capable artificial intelligence systems.
The business informed Bloomberg News that OpenAI is now more extensively integrating the so-called superalignment team across its research activities to help the company reach its safety goals, as opposed to keeping the group as a stand-alone entity. Jan Leike, another OpenAI veteran, and Sutskever led the team when it was founded less than a year ago.
The decision to reevaluate the team was made in response to concerns raised by a number of recent OpenAI departures regarding the company’s strategy for striking a balance between speed and safety when creating its AI products. Respected researcher Sutskever stated on Tuesday that he was leaving OpenAI. Sutskever and CEO Sam Altman had previously disagreed over how quickly artificial intelligence should advance.
Leike posted a curt message on social media shortly afterward to announce his departure. He declared, “I resigned.” Sutskever’s departure was the final straw for Leike after conflicts with the corporation, according to a person with knowledge of the matter who wished to remain anonymous to discuss private discussions.
Leike claimed that the superalignment team had been battling for resources in a statement on Friday. Leike posted on X, “My team has been sailing against the wind for the past few months.” “There were moments when we were short on processing power and it was becoming more difficult to complete this important research. “Altman replied to Leike’s post a few hours later. Altman posted on X, “He’s right, we still have a lot of work to do.” “We’re determined to complete it.”
In recent months, a few more members of the superalignment team have also left the organization. OpenAI fired Leopold Aschenbrenner and Pavel Izmailov. Their departures were previously revealed by The Information. A person with knowledge of the situation claims that Izmailov had already been moved off the team before leaving. Requests for feedback from Izmailov and Aschenbrenner were not answered.
The startup’s co-founder John Schulman, whose work focuses on massive language models, will head OpenAI’s alignment efforts moving forward, the firm announced. In a separate blog post, OpenAI announced that Jakub Pachocki, Research Director, will succeed Sutskever as Chief Scientist.
Regarding Pachocki’s hiring, Altman said in a statement on Tuesday, “I am very confident he will lead us to make rapid and safe progress towards our mission of ensuring that AGI benefits everyone.” Artificial general intelligence, or AGI, is defined as AI that can accomplish most jobs just as well as or better than humans. Though it doesn’t currently exist, one of the company’s goals is to develop AGI. Employees at OpenAI are working on individual safety-focused teams as well as company-wide teams linked to AI safety. One, a readiness team, was established in October of last year with the goal of examining and mitigating the “catastrophic risks” associated with AI systems.
The goal of the superalignment team was to neutralize the biggest long-term dangers. The superalignment team was formed by OpenAI, which had long declared its technological objective to regulate and guarantee the security of artificial intelligence software that will be smarter than humans. The team was formed in July of last year. OpenAI stated in the announcement that the team’s work will receive 20% of its processing power at that time.
Sutskever was one of several OpenAI board members who pushed for Altman’s termination in November, a move that set off a hectic five days for the business. After OpenAI President Greg Brockman resigned in protest, investors staged a mutiny, and within a few days, almost all 770 workers of the business signed a letter threatening to depart if Altman was not reinstated. Remarkably, Sutskever also signed the letter and expressed regret for his role in Altman’s dismissal. Altman was back on board shortly after.
Sutskever mainly vanished from the public eye in the months that followed Altman’s departure and return, raising questions about his potential role at the corporation. According to a person with knowledge of the situation, Sutskever also ceased working from OpenAI’s San Francisco headquarters.
Leike claimed in his statement that he left OpenAI due to a number of differences with the company’s “core priorities,” which he believes are not sufficiently concentrated on safety precautions associated with the development of AI that may be more intelligent than humans.
Sutskever stated he is “confident” OpenAI will create AGI “that is both safe and beneficial” under its present leadership, which includes Altman, in a post earlier this week announcing his departure.
