Pages

26 November 2023

What is Project Q*? OpenAI board was warned of 'dangerous' AI discovery before Altman ouster


In a fresh revelation, it has now surfaced that several researchers at OpenAI wrote to the board of directors warning of a powerful AI discovery that could threaten humanity, said a report, citing sources.

The letter and the discovery of the AI algorithm were the major developments at the San Francisco-based startup before the board pushed out Sam Altman on Friday, Reuters reported, quoting two sources familiar with the matter.

Altman was reinstated as OpenAI CEO on Wednesday after pressure from the company's biggest financial backer Microsoft and its CEO Satya Nadella as well as warning from over 700 employees who threatened to quit in support of the fired chief.

Talking to the news agency, the sources said the letter was one important factor that prompted the board of director to fire Altman amid concerns over commercialising the AI discovery without gauging the aftereffects.

Project Q*

OpenAI, in an internal message to staffers, acknowledged a project called Q* (pronounced Q-Star) and a letter to the board before the weekend's events, one of the people said.

An OpenAI spokesperson said that the message, sent by long-time executive Mira Murati, alerted staff to certain media stories without commenting on their accuracy. Murati, who was appointed interim CEO on Friday after Altman's firing, quit OpenAI two days later after initial talks to bring back the ousted chief failed.

A section of OpenAI staff believe that Project Q* could be a breakthrough in OpenAI's search for artificial general intelligence (AGI), according to a source cited by Reuters. According to the company, AGI are “autonomous systems that surpass humans in most economically valuable tasks.”

The AI algorithm could ace grade-school math and researchers at OpenAI were optimistic about the future success of Q*, the source told Reuters.

Generative AI is currently good at writing and language translation where the answers can vary widely. However, the ability to solve math problems where there is only one correct answer indicates that AI is capable of performing close to the level of human intelligence, according to the new agency.

In the letter sent to the OpenAI board, staff researchers warned of this powerful algorithm's potential danger to humanity, the sources were quoted as saying. However, the specific safety concerns noted in the letter were not revealed.

No comments:

Post a Comment