OpenAI’s Math Claims Draw Scientific Pushback

NPR’s “Machine Gods” episode, published October 30, 2026, reports that OpenAI says its AI solved three difficult problems in mathematics, computer science and physics. The claims have angered scientists, who are challenging how the company disclosed the results.

Researchers’ objections include proofs they say are difficult to verify and OpenAI selecting who gets to review them first. The episode examines the company’s claims, changes to its release process and questions about the role of human scientists if AI continues solving problems that stumped experts for decades.

The episode also begins “Humans in the Loop,” a listener mailbag. It addresses what a parent might tell a teenager who considers AI evil when the parent’s job requires using it, how to stop a rogue AI agent swarm, and whether it is acceptable to secretly change a partner’s ChatGPT settings.

The episode runs 47:39 and is hosted by Kevin Roose and Casey Newton, with Danush Parvaneh, Rachel Cohn, Jacob Blumberg and Sasha Ashall also credited. NPR says it does not offer or accept money for coverage or interviews.

NPR’s “Machine Gods” episode, published October 30, 2026, reports that OpenAI says its AI solved three difficult problems in mathematics, computer science and physics. The claims have angered scientists, who are challenging how the company disclosed the results.

Researchers’ objections include proofs they say are difficult to verify and OpenAI selecting who gets to review them first. The episode examines the company’s claims, changes to its release process and questions about the role of human scientists if AI continues solving problems that stumped experts for decades.

The episode also begins “Humans in the Loop,” a listener mailbag. It addresses what a parent might tell a teenager who considers AI evil when the parent’s job requires using it, how to stop a rogue AI agent swarm, and whether it is acceptable to secretly change a partner’s ChatGPT settings.

The episode runs 47:39 and is hosted by Kevin Roose and Casey Newton, with Danush Parvaneh, Rachel Cohn, Jacob Blumberg and Sasha Ashall also credited. NPR says it does not offer or accept money for coverage or interviews.