An international research team involving Korean mathematicians has produced meaningful solutions related to 13 world-class unsolved mathematical problems by using Google DeepMind’s mathematical artificial intelligence (AI) agent “Aletheia.”

According to the mathematics community on the 8th, Aletheia reviewed about 700 problems from the “Erdős problems,” a collection of unsolved mathematical problems, and presented meaningful solutions to 13 of them. The research team, which includes Professor Kim Sang-hyun of the Korea Institute for Advanced Study (KIAS), Professor Jung Junhyuk of Brown University in the United States, Senior Researcher Tang Luong of Google DeepMind, and Professor Tony Feng of the University of California, Berkeley, published a paper containing these findings on the preprint server “arXiv” on the 2nd.

The Erdős problems are a collection of problems in combinatorics and number theory posed by Hungarian mathematician Paul Erdős. Of a total of 1,179 problems, about 700 remain unsolved. In recent years, they have been used as a representative benchmark for assessing AI’s mathematical capabilities.

The research team fed Aletheia 700 Erdős problems over about one week in December last year. Aletheia, which means “truth” in Ancient Greek, is a mathematical research agent based on “DeepSeek,” the advanced reasoning mode of Gemini, Google’s generative AI. According to the research team, Aletheia reviewed approximately 700 unsolved problems in a short period and derived meaningful solutions for 13 of them. In the process, it exchanged proofs with human mathematicians and expanded theories, demonstrating a collaborative style similar to that of an actual mathematician.

Professor Kim Sang-hyun said, “AI and mathematicians exchanged ideas and produced results at the level of professional researchers,” adding, “On the basis of the order built up by the mathematical community over thousands of years, using AI will be beneficial for humanity.”

Popular News