Engineer Ilya Sutskever was one of the engineers who called for the dismissal of Sam Altman last year due to his commercial drift. He left OpenAI, the company he helped found, last May.
Among the original founders of OpenAI, the company behind the currently most popular artificial intelligence application, ChatGPT, it is hard to find someone with greater moral conviction than Ilya Sutskever.
This Canadian-Israeli engineer (he is also a Russian citizen) was until May the chief scientist of the company. He contributed to launching it as a nonprofit organization eight years ago, and his vision has remained unchanged since: we are on the brink of creating what is known as Artificial General Intelligence, an AI capable of reasoning and learning like a human, and it is too important a discovery to be left in the hands of a company like Google or another tech giant.
Sutskever was also one of the voices that managed to convince OpenAI’s board of directors of the need to dismiss its then-CEO, Sam Altman, last November. Altman spent four days in professional limbo before returning to the leadership of OpenAI after a fierce internal battle. Among those calling for his removal, including Sutskever, the main concern was the decision to forge a commercial agreement with Microsoft and the lack of concern for the ethical implications of the company’s developments.
Sutskever’s departure last May, therefore, did not catch anyone by surprise. Although Altman and Sutskever have always maintained a good public relationship, they represent two very different visions of artificial intelligence. Now, with his power strengthened, Altman is also trying to erase any trace of OpenAI’s past as a nonprofit organization and transform the company into a conventional Silicon Valley enterprise.
The remaining question was what this engineer, who had previously also worked for Google, would do outside of OpenAI. Yesterday, Sutskever announced that his next project will be to create safe superintelligence. “It is the most important project of the era we live in,” he explains on the website of his new company, which he has named Safe Superintelligence Inc. (SSI) and which will be based in California and Tel Aviv.
Sutskever will be joined by former OpenAI engineer Daniel Levy and Daniel Gross, co-founder of Cue and former head of artificial intelligence at Apple. So far, the company has not explained what its product will be or how it will secure funding, although, in statements to Bloomberg, Gross has stated that they will not have trouble attracting capital.
SSI could face the same challenges that have forced OpenAI to change its strategy and philosophy. Developing language models like GPT-4 (the engine behind ChatGPT) requires enormous processing power. This is why OpenAI had to reach an agreement with Microsoft, which provides the company with the powerful data centers it needs, and also the reason Altman wants to abandon OpenAI’s current model (now a for-profit entity with limited return on investment and subject to the control of a central nonprofit organization) and shift to a conventional company model that facilitates seeking investors and, later, potentially going public.
