In a secret showdown, 30 top mathematicians tried to stump OpenAI's new AI, o4-mini. With $7,500 on the line for each failed problem, the AI's shocking success left them speechless and questioning the future.
May 2025. 30 elite mathematicians met in secret to challenge a new AI with "unsolvable" problems.
The challenger: OpenAI's o4-mini. A new AI designed to solve math problems beyond human genius.
The prize? $7,500 for every question the AI couldn't answer. The ultimate human vs. machine challenge.
The result was stunning. The experts tried their best, but only 10 of their questions stumped the bot.
One of math's biggest unsolved problems was cracked by the AI in under 10 minutes, with a "cheeky" solution.
A new fear emerged: "proof by intimidation." Can we trust an AI that sounds more confident than we are?
The role of mathematicians may be changing forever—from problem solvers to AI collaborators and question curators.