Summary
Highlights
The video discusses a newly released 62-page deposition from Ilya Sutskever, former OpenAI founder, as part of Elon Musk's lawsuit against Sam Altman. Musk's lawsuit alleges that Altman illegally converted OpenAI from a non-profit to a for-profit entity, betraying its original mission. The deposition focuses on Sutskever's efforts to fire Altman and the internal dynamics of OpenAI's board.
Ilya Sutskever prepared a 52-page memo to build a case for firing Sam Altman. He accused Altman of a consistent pattern of lying, undermining executives, and pitting them against each other. Sutskever shared this memo only with independent board members, fearing Altman would make the discussions 'disappear' if he became aware of them. Mira Murati, former CTO, also contributed evidence, including screenshots, to support Sutskever's claims.
Ilya Sutskever collaborated with independent board members like Adam D'Angelo (Quora CEO, still on OpenAI's board), Helen Toner (Australian AI researcher), and Tasha MacAulay (tech entrepreneur) to build the case against Altman. Helen Toner was particularly vocal, even criticizing OpenAI in an article while still on its board, and believed that OpenAI's destruction would be consistent with its mission if Altman remained.
Sutskever believed Altman's termination was the appropriate action. He even sent the memo using disappearing emails due to concerns about leaks. Despite the board's initial decision to fire Altman, he was reinstated within a week, and most of original board resigned. Sutskever believed the firing process was rushed due to the board's inexperience, particularly Helen Toner and Tasha MacAulay's lack of familiarity with OpenAI's day-to-day operations and products like ChatGPT.
Murati informed Sutskever that Altman was allegedly removed from Y Combinator for similar reasons (creating chaos, pitting people against each other, poor management), a claim that Paul Graham, YC co-founder, refuted by stating Altman left to focus on OpenAI. Murati also alleged that Greg Brockman was fired from Stripe, another claim Sutskever did not verify. She further accused Altman of pitting Daniela Amodei (Dario Amodei's sister) against her. Sutskever admits he didn’t verify these claims.
Dario Amodei, founder of Anthropic, aimed to become the head of research at OpenAI, replacing Greg Brockman. Sutskever faulted Altman for not making a firm decision on Amodei's demands, which Sutskever deemed unfair. Altman was also accused of telling Sutskever and Jakob Pachocki (current chief scientist at OpenAI) conflicting things about running the company.
During the conflict surrounding Altman's firing, Anthropic attempted to merge with OpenAI, with its leadership taking over. Helen Toner was highly supportive of this proposal, but Sutskever was strongly against it. Sutskever revealed he had considered Altman's removal for at least a year prior to the actual attempt, though he hadn't planned it actively due to perceiving it as unfeasible.
Sutskever was surprised by OpenAI employees' strong negative reaction to Altman's firing, expecting a more neutral response. He later left OpenAI to start his own frontier AI company, Safe Super Intelligence, due to a new vision. He confirmed that his equity in OpenAI increased in value after his departure, and stated that he is unsure who is paying his legal fees for the current lawsuit, suspecting it might be OpenAI itself.