It is a problem formulated in 1946 by the Hungarian mathematician Paul Erdős, and concerns how many equal-length lines can be drawn between points on a large sheet of paper.
Many mathematicians have worked on the problem. Erdős's own theory was that the most lines can be drawn if the points lie in a grid. A model from OpenAI has now shown that Erdős's assumption was completely wrong.
It found tools in another part of mathematics, Linusson tells DN.
The AI also wrote a research article about the solution.
If a human had done it, the article could have been published in the most prestigious journals, and the person would have been a strong candidate for a job at most universities in the world.
