OpenAI disbanded its dedicated AI risk mitigation team in less than a year

OpenAI has broken it down Super Alignment Team, dedicated to mitigating the long-term risks of artificial intelligence (AI). This event takes place just one week after its launch GPT-4oyour model Amnesty International The most advanced yet. As mentioned WiredThis team was created in July 2023 and was led by Ilya Sutskever And Jan Lake.

Team solution Super alignment This came a few days after its leaders Sutskiver And LikeThey will announce their resignations on social media. In his statement, Sutskever expressed confidence that “OpenAI will build safe and useful AGI” under current leadership. He added that he is excited about a new project that he considers “of personal importance,” although he did not provide further details about it.

For his part, he explained to you: s His departure was due to disagreements over the company’s core priorities, stating that safety and operations had been neglected “by flashy products.” He noted that his team was struggling to obtain critical computational resources for their research, making it more difficult to accomplish their mission.

As shown CNBCthe team Super alignment It is tasked with solving the fundamental technical challenges of superintelligence alignment within four years, using 20% ​​of computing power OpenAI. The risks they addressed included technology misuse, economic disruption, misinformation, bias and discrimination, addiction and over-reliance.

Following resignations and dissolution of the team, some members were reassigned to other teams within the team OpenAI. NBC News These transfers were confirmed by an anonymous source. However, when the company was asked to comment, it only recommended the latest publication Sam Altman in sHe expressed his sadness at the departure of Likee and indicated that the company has more work to do.

The departure of Sutskever and Leike reminds us of the leadership crisis he went through OpenAI Last November, when the company’s board of directors was dismissed AltmanThis sparked a chain of reactions that included resignations and threats of resignation by employees and investors, including… Microsoft.

Sutskever, who played a key role in Altman’s ouster, later expressed his regret for his participation in the event.

Although the aforementioned manager stayed on OpenAI As chief scientist, he no longer holds a seat on the board. After announcing his departure, Altman commented on this s: “This is very sad for me; “Elijah is easily one of the brightest minds of our generation, a guiding light in our field, and a dear friend.” Likewise, a CEO was appointed Jacob Paczucki His successor as Chief Scientist.

Interested in trade He pointed out that there was news of the disbanding of the team Super alignment The resignation of Sutskever and Leike coincided with the launch of new products before OpenAI. The company has provided an update to the model GPT-4 And the desktop version of ChatGPTWith an improved user interface and new video chat capabilities.