New AI model disproves complex mathematical conjecture

According to OpenAI, the company’s new reasoning model has presented an independent mathematical proof that refutes a long-standing and complex hypothesis in geometry, based on publicly available data. This was reported by Zamin.uz.

The scientific problem in question was first proposed in 1946 by the renowned mathematician Paul Erdős, and over the decades, many scholars have struggled to find its solution. Although OpenAI has made similar claims in the past, this latest result is being taken seriously and substantively by the scientific community.

A few months ago, company representatives announced that new models had found answers to several of Erdős’s problems. However, it later became clear that the system had simply reprocessed existing information from the literature, which drew sharp criticism from experts in the field and leaders of competing companies.

This time, the OpenAI team learned from past mistakes, subjected the results to rigorous verification in collaboration with world-renowned mathematicians, and confirmed their validity. For nearly sixty years, mathematicians had believed that the most popular solution to this problem relied on geometric shapes of a specific order.

But the new model has completely overturned this conventional view, discovering a new class of constructions that operate even more efficiently. This achievement demonstrates that artificial intelligence has reached a new level in its ability to independently solve complex logical problems.

As emphasized by the company’s specialists, this breakthrough is not limited to mathematics alone. The system in question was designed not for narrow computations, but for broad, general-purpose reasoning.

Such high-level intellectual capability is expected to serve humanity in the future by enabling new, previously unknown discoveries in vital fields such as biology, physics, engineering, and medicine.