OpenAI claims to have solved 377 math problems, but Danish professor is unimpressed

OpenAI says it has produced solutions to 377 mathematical problems that researchers have struggled with for years. According to Børsen, the company presented the results on Tuesday, and the news hit the world's leading mathematicians hard. These are problems that human mathematicians have spent years on, often standing frustrated in front of large blackboards while trying to crack the equations.

The article says each of the 377 results required computing power equivalent to roughly three hours of thinking in ChatGPT Pro, on average.

A Danish professor is not impressed. The article's standfirst says he calls the result only a "half breakthrough" and criticises OpenAI for setting aside the understanding of the mathematics itself. The text available here is mostly the headline, standfirst and opening paragraphs, plus unrelated headlines from the Børsen front page. It does not include the professor's name, his full argument, or details of which problems were solved or how the solutions were checked.

The piece matters because it sets a large claim about AI's mathematical ability against a Danish expert's scepticism. The criticism is that producing answers is not the same as understanding the mathematics behind them. The article was written by William Kring Østergaard and published on 9 October 2026.