Semafor Signals
OpenAI forms safety council — while training a powerful new AI model
Insights from The Economist, Financial Times, and The Information
The News
ChatGPT creator OpenAI is training a powerful new model to fuel its chatbot and image generation tools, the company said Tuesday. It is also launching a new committee focused on safety, following scrutiny over its safety efforts and several high-profile resignations.
The moves follow a controversy earlier this month in which OpenAI suspended a voice chatbot after actress Scarlett Johansson accused the company of copying her AI voice character from the movie Her.
OpenAI said the next model will “bring us to the next level of capabilities on our path to AGI,” which refers to artificial general intelligence — AI systems that could eventually match or surpass human capabilities.
SIGNALS
Former OpenAI board members skeptical of self-governance
Following the recent staff revolt at OpenAI — in which one former executive said safety took a backseat to “shiny products” — the new safety committee will be led by CEO Sam Altman himself, along with two of the company’s board members. Two former board members, though, argued in The Economist that “self-governance cannot reliably withstand the pressure of profit incentives.” They specifically called out Altman, saying he “undermined the board’s oversight of key decisions and internal safety protocols.” The board members urged lawmakers to take the lead on AI regulation, instead of letting big tech companies govern themselves.
OpenAI backtracks on superintelligence efforts
OpenAI appears to be sending mixed signals about its mission around superintelligence, which refers to AI systems that are smarter than humans. CEO Sam Altman said last year that he spends half of this time researching “how to build superintelligence,” but a top OpenAI executive backtracked on those claims, telling the Financial Times that the company’s “mission” is to build AGI, not superintelligence. A spokesperson said the company studies superintelligence as part of its artificial general intelligence mission. Earlier this month, OpenAI dissolved its “Superalignment” team less than a year after it vowed to use 20% of its computing power on the team’s efforts to control the long-term risks associated with AI and superintelligence.
Company faces competition from Musk
OpenAI is facing intensified competition from Elon Musk’s artificial intelligence startup xAI that just raised $6 billion from big-name investors. The funds will be used to build infrastructure, drive research and development, and bring products to market, xAI said. Musk is trying to lead the way on superintelligence development, and part of his plans include building a new supercomputer to power the next version of its chatbot, according to The Information. While xAI is a relative newcomer, it has made major hires and Musk himself was among the founders of OpenAI — though he has since had a fractious relationship with it.
