A recent scientific experiment has revealed distinctive capabilities in which humans still outperform artificial intelligence, as AI models failed to surpass mathematics experts during a complex research test conducted under the "First Proof" project — one of the most difficult mathematics tests to date — in which four AI systems were asked to solve ten complex research problems.
The problems were not included in the training data of the participating models, and specialist mathematicians reviewed and evaluated the answers.
The results showed, according to Russia Today, that current AI models remain less capable than senior mathematicians when dealing with similar problems, and that they also lack mathematical intuition and remain prone to making errors, or what is known as "hallucination."
Ten researchers proposed these problems from their own unpublished scientific work.