Google Deepmind's AlphaProof Nexus Cracks Decades-Old Math Problems on a Shoestring

In a groundbreaking achievement, Google Deepmind's AlphaProof Nexus has autonomously solved nine open Erdős problems, including two that had confounded mathematicians for an astonishing 56 years. The cost of achieving this feat? A mere few hundred dollars per problem in inference costs.

This remarkable breakthrough showcases the potential of artificial intelligence in tackling some of the world's most enduring mathematical challenges. The approach taken by AlphaProof Nexus differs significantly from that of OpenAI, which has focused on natural-language processing. Instead, AlphaProof Nexus utilizes the Lean compiler to automatically verify every proof step, ensuring a high degree of accuracy and rigor in its solutions.

This innovative methodology has enabled the system to make significant inroads into the realm of mathematical problem-solving. Despite this impressive achievement, it's worth noting that the overall success rate of AlphaProof Nexus stands at a relatively modest 2.5 percent. This suggests that there is still much work to be done in refining the system and improving its ability to tackle complex mathematical problems.

Nevertheless, the progress made by AlphaProof Nexus represents a significant step forward in the field of AI-assisted mathematics. The implications of this breakthrough are substantial, and mathematicians are likely to be closely watching further developments in this area. As AI systems like AlphaProof Nexus continue to advance, they may unlock new insights and solutions to some of the world's most pressing mathematical challenges.

For now, the achievement of solving nine open Erdős problems, including two that had resisted solution for over five decades, stands as a testament to the power of AI in tackling complex problems. The article originally appeared on The Decoder, a platform that tracks the latest developments in AI and related technologies.