The ouster of Sam Altman from OpenAI last week was reportedly influenced by concerns about a potentially major artificial intelligence breakthrough achieved by the company.

Reuters reported the twist late Wednesday, citing two people familiar with the matter. The sources said that some OpenAI researchers had expressed concerns the breakthrough “could threaten humanity.” It’s believed the researchers sent a letter about the matter to OpenAI’s board shortly before it ousted Altman.

Whether such a document was indeed penned is unclear. This morning, The Verge cited a source as saying that OpenAI’s board had never received the letter described in the Reuters report. 

Altman was ousted from OpenAI last Friday following disagreements with members of the AI developer’s board. In response, most of the company’s 770 employees signed a letter saying they would resign unless the executive returned. On Tuesday, OpenAI announced that it has “reached an agreement in principle” to reappoint Altman as CEO and reorganize the board.

According to Reuters, a group of OpenAI researchers wrote a letter to the board before Altman’s ouster. In the letter, they reportedly highlighted the potential risks that could be posed by advanced AI. They drew particular attention to an internal OpenAI project dubbed Q*, pronounced Q-Star. 

According to the report, some at the company believe Q* could represent a major breakthrough in the journey toward building artificial general intelligence, or AGI. That’s the term for a hypothetical future AI system with human-level reasoning skills. The report described the breakthrough as a new “AI algorithm” without going into much detail.

The algorithm repeatedly demonstrated the ability to solve grade school math problems with a high level of accuracy. There are already AI systems on the market that can solve such problems, but it appears Q* carries out the task with a different, more sophisticated approach. Internal tests of the AI reportedly made OpenAI researchers “very optimistic about Q*’s future success.”

According to Reuters, the letter to OpenAI’s board in which the researchers expressed their concerns about Q* also flagged the work of an internal group called the “AI scientist team.” This team reportedly focuses on finding ways to improve AI models’ reasoning skills. It’s said to have been formed through the merger of two earlier OpenAI groups that focused on enhancing neural networks’ math and coding know-how.

OpenAI executive Mira Murati reportedly acknowledged the existence of a project called Q* in an internal memo sent following this week’s report. However, the memo didn’t comment on the reports’ accuracy. “Mira told employees what the media reports were about but she did not comment on the accuracy of the information,” OpenAI told The Verge in a statement.

Reuters described the concern over Q* as only one of several factors that influenced the board’s decision to oust Altman. According to the report, the directors involved in the decision were also worried OpenAI is proceeding too fast with its AI commercialization efforts. 

The news that Altman will return to OpenAI appears to have boosted investor confidence in the company’s prospects. According to The Information, a planned secondary sale of OpenAI shares at a $86 billion valuation was paused following the executive’s ouster. The transaction is now back on track, the publication reported today.

A group of investors led by Thrive Global is expected to purchase up to $1 billion worth of OpenAI shares from existing stakeholders. Moreover, there’s reportedly a chance the value of the shares sold could exceed $1 billion. The transaction is expected to close next month. 

Photo: Unsplash