Sam Altman says AI might one day solve questions even he can’t grasp

In a conversation that veered from light-hearted to deeply reflective, OpenAI CEO Sam Altman shared an anecdote about GPT-5’s capabilities that left him both impressed and unsettled.

Speaking on comedian Theo Von’s 'This Past Weekend' podcast, Altman recounted a moment where he used GPT-5 to help answer a question he personally didn’t understand. “I put it in the model, this is GPT-5, and it answered it perfectly,” Altman said. “It was a weird feeling… I felt useless relative to the AI.”

The moment, he explained, was a revealing snapshot of just how powerful the latest iteration of OpenAI’s generative model has become. While GPT-5 has not yet been officially released, Altman’s comments suggest that it represents a meaningful leap in reasoning ability. “It integrates a lot of our technology,” he added, referring to the model’s ability to unify several of OpenAI’s recent advancements under one system.

Altman did not provide a definitive release date for GPT-5, but his remarks have added momentum to industry speculation that a launch may be imminent. His interaction with the model has also reignited broader conversations about how generative AI could shift traditional expectations of human knowledge and cognitive limits.

The interview comes at a time when OpenAI faces heightened competition, internal restructuring, and philosophical debates over how close we are to achieving artificial general intelligence. For Altman, though, the personal impact of AI’s advancement is already clear: “I felt like I should have been able to answer the question, but GPT-5 answered it instantly.”