Google DeepMind's AlphaProof Nexus AI Solves 56-Year-Old Math Problems

Google DeepMind's AlphaProof Nexus AI system autonomously solved nine open Erdős problems, some unsolved for nearly 56 years, and proved 44 open OEIS conjectures, announced May 26, 2026. The system also solved a 15-year old algebraic geometry question and discovered a new algorithmic parameter in optimization theory, using only a few hundred dollars worth of computing power per problem. The AI generated and verified proofs using formal logic with minimal human intervention.

AlphaProof Nexus mathematical achievements

• ▪Google DeepMind's AlphaProof Nexus solved each problem using only a few hundred dollars worth of computing power.
• ▪Google DeepMind's AlphaProof Nexus discovered a completely new algorithmic parameter in optimization theory that had not been identified before.
• ▪The article was published on May 26, 2026.
• ▪OpenAI claimed that its AI models solved the famous planar unit distance problem, which was first proposed by mathematician Paul Erdős in 1946.
• ▪Google DeepMind's AlphaProof Nexus AI system autonomously solved nine open Erdős problems, some of which remained unsolved for nearly 56 years.
• ▪Google DeepMind's AlphaProof Nexus reportedly proved 44 open OEIS conjectures.
• ▪Google DeepMind's AlphaProof Nexus solved a 15-year old algebraic geometry question.

Autonomous proof generation methodology

• ▪Google DeepMind's AlphaProof Nexus generated mathematical proofs on its own and verified them using computer based formal logic.
• ▪Google DeepMind's AlphaProof Nexus generated proof attempts autonomously along with continuously checking its own reasoning using formal mathematical verification tools.
• ▪Google DeepMind's AlphaProof Nexus worked with minimal human intervention.

AGI limitations debate

• ▪Demis Hassabis argued that genuine AGI would need to be original, creative and have a wide range of skills across a multitude of areas, not just a specific one.
• ▪Demis Hassabis stated that no AI system can be considered as having human intelligence just by solving complex equations.
• ▪Demis Hassabis stated that today's AI systems are nowhere near achieving true AGI despite recent advances in mathematical reasoning.
• ▪Demis Hassabis likened modern AI systems with great mathematicians like Srinivasa Ramanujan and stated that today's AI systems are still far behind the depth of intuition and innovation displayed by extraordinary human minds.

Perspective of Demis Hassabis (Google DeepMind CEO)

• ▪Demis Hassabis argued that genuine AGI would need to be original, creative and have a wide range of skills across a multitude of areas, not just a specific one.
• ▪Demis Hassabis stated that today's AI systems are nowhere near achieving true AGI despite recent advances in mathematical reasoning.

1 source