OpenAI has dissolved its superalignment team according to a person who works at the company. The superalignment team was set up last year with the job of finding a way to keep human control over an AI that eventually becomes smarter than us.
Some of the team members who were working on this have been moved to several other teams throughout the company. The news comes after senior OpenAI members Ilya Sutskever and Jan Leike left the company.
Leike announced his resignation on Twitter and said that the safety culture at the company had taken a backseat with it becoming more of a struggle for the team to get compute resources to do their work.
Over the past few months my team has been sailing against the wind. Sometimes we were struggling for compute and it was getting harder and harder to get this crucial research done.
— Jan Leike (@janleike) May 17, 2024
Following Leike’s resignation, Sam Altman, head of OpenAI, said that Leike is right about safety and that OpenAI is committed to safety going forward. Altman said that he will write a longer post on the matter in the next couple of days.
OpenAI is a pretty opaque company when it comes to sharing development progress so it’s not really clear how far away we are from this super intelligent AI. If the company is coming up against any roadblocks due to things like limitations on compute power, then perhaps Altman doesn’t see the need for a superalignment team just yet – we will simply have to wait for an explanation in the coming days.
i'm super appreciative of @janleike's contributions to openai's alignment research and safety culture, and very sad to see him leave. he's right we have a lot more to do; we are committed to doing it. i'll have a longer post in the next couple of days.
— Sam Altman (@sama) May 17, 2024
🧡 https://t.co/t2yexKtQEk
The last week has been a massive week for the company. It told the world about its latest model, GPT-4o which is available to free and premium users in a limited capacity before falling back to GPT-3.5 for a few hours, ensuring all users can have a go with it.
It also showed off new vision and voice capabilities of the AI which are now native to GPT. With them, the model can change its voice tone at your request and it can also see the world around you if you allow it to.
Source: Wired via CNBC | Image via Depositphotos.com
3 Comments - Add comment