OpenAI's AI reasoning expert Noam Brown says there is "lots of vague AI hype" on social media.
While acknowledging there are "good reasons to be optimistic" about AI progress, Brown emphasized that plenty of research problems still need solutions, adding that Open has "not yet achieved superintelligence."
His comments directly challenge recent statements from within OpenAI. In January, OpenAI researcher Stephen McAleer wrote: "I kinda miss doing AI research back when we didn't know how to create superintelligence," suggesting that OpenAI at least found a clear path to so-called ASI (artificial super intelligence).
Before joining OpenAI, Brown worked at Facebook AI Research (FAIR), where he developed AI systems that defeated human players in complex games like poker and Diplomacy. His work on systems like the poker AI Libratus explored the concept of "test-time compute," showing that giving AI more calculation time led to better game moves.
Brown later brought these concepts to OpenAI, applying them to language models. The company's recent o1 model is a direct result of this work, taking a different approach by scaling "thinking time" rather than just increasing training performance.
Brown believes this alternative scaling method could enable new AI capabilities. "We're still early in scaling along that dimension," Brown writes.
