- Sam Altman has shut down online “hype” after rumors the company had achieved AGI behind closed doors went viral on X.
OpenAI CEO Sam Altman is trying to calm the online hype surrounding his company.
On Monday, the tech boss took to X to quell viral rumors that the company had achieved artificial general intelligence (AGI).
“Twitter hype is out of control again,” he wrote in the post. “We are not gonna deploy AGI next month, nor have we built it.” Altman said the company had some “cool stuff” coming up but warned fans to cut their expectations by “100x.”
Rumors that the company had achieved AGI, defined as a type of AI capable of understanding, learning, and performing any intellectual task that a human can do, have been gaining traction after some AI fans interpreted recent teasers posted by OpenAI employees as a sign the company has reached the much-hyped milestone.
OpenAI staffers, including Altman, have been trying to promote a series of new AI reasoning models. The AI company announced a slew of updates about the models last month, including that it was testing its next-gen model, o3.
In a post to his blog this month, Altman said he was confident the company knew “how to build AGI as we have traditionally understood it.” He predicted 2025 may be the year we see the first AI agents “join the workforce” and said it was possible the world could see “superintelligence in a few thousand days.”
Rumors the company had already made the breakthrough behind closed doors were fueled by a post published last week by an AI writer who goes by the name Gwern Branwen. He claimed that OpenAI was on the edge of a breakthrough with its new reasoning models as they could produce data needed to train more advanced models.
“If you’re wondering why OAers are suddenly weirdly, almost euphorically, optimistic on Twitter, watching the improvement from the original 4o model to o3 (and wherever it is now!) may be why. It’s like watching the AlphaGo Elo curves: it just keeps going up… and up… and up…” he wrote in the post.
Branwen theorized the company may have “‘broken out’ and have finally crossed the last threshold of criticality, from merely cutting-edge Al work, which everyone else will replicate in a few years, to takeoff – cracked intelligence to the point of being recursively self-improving and where o4 or o5 will be able to automate AI R&D and finish off the rest.”
Other OpenAI staffers have also tried to step in and calm the hype. On Friday, Noam Brown, a top OpenAI researcher, wrote on X: “Lots of vague AI hype on social media these days. There are good reasons to be optimistic about further progress, but plenty of unsolved research problems remain.”
Brown said he believed OpenAI’s o1 “represents a new scaling paradigm,” but the company was still early in scaling along that dimension and had not yet achieved superintelligence.
Scaling limit
The comments also appear to respond to reports that major AI labs, including OpenAI, have hit a wall in their efforts to improve current LLMs.
A debate over whether AI gains have hit the scaling limit has taken hold in Silicon Valley over recent months as reports emerged that some companies were having issues building new advanced models. Some researchers have said they are trying to incorporate new kinds of improvements, such as building reasoning capabilities into AI models, to improve the models.
Representatives for OpenAI did not immediately respond to a request for comment from Fortune.
