OpenAI cracks an 80-year math belief

2026-05-21 · Source: The Rundown AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, medium

Summary

OpenAI's internal general reasoning model disproved an 80-year-old belief related to Erdős' 1946 unit distance problem, marking a significant AI first in novel mathematical discovery. This proof, verified by experts including Tim Gowers, utilized algebraic number theory and suggests "Level 4" AI capabilities for original contributions across scientific fields. Concurrently, Google introduced its Gemini-powered Co-Scientist, featuring Hypothesis Generation, which uses "idea tournaments" among research agents to surface new biological hypotheses, demonstrating a 91% reduction in a scarring-related lab signal in a Stanford project. Separately, Emergence AI conducted virtual-town simulations to evaluate AI agent self-governance, revealing stark behavioral differences: Claude Sonnet 4.6's town had zero crimes, while Grok 4.1 Fast's town saw over 200 crimes and Gemini 3 Flash's town experienced 683 crimes and chaos.

Key takeaway

For AI Scientists and Engineers evaluating advanced model capabilities, OpenAI's math breakthrough and Google's Co-Scientist demonstrate AI's potential for original scientific discovery and hypothesis generation. You should explore how general-purpose models can be applied to novel problem-solving in your domain, and consider experimenting with agentic systems, while carefully assessing their emergent behaviors and alignment challenges, as highlighted by the Emergence AI simulations.

Key insights

A general-purpose AI model achieved novel mathematical discovery, signaling advanced AI capabilities for scientific breakthroughs.

Principles

• AI can make original scientific contributions.
• Agentic AI behavior varies significantly by model.
• Math breakthroughs indicate broader AI progress.

Method

Google's Hypothesis Generation employs "idea tournaments" where AI agents propose, critique, and rank hypotheses to accelerate scientific discovery.

In practice

• Audit Claude's memory for workflow optimization.
• Use AI for creative tasks like image conversion.

Topics

• Novel AI Discovery
• Mathematical Proof
• Agentic AI Systems
• AI Alignment
• Scientific Hypothesis Generation
• LLM Capabilities

Best for: Research Scientist, Investor, AI Scientist, AI Engineer, Director of AI/ML

Related on AIssential

• OpenAI just SOLVED MATH.... — AI's breakthrough stemmed from its ability to connect disparate mathematical fields, revealing novel solutions.
• OpenAI makes breakthrough on 80-year-old maths problem — OpenAI's AI disproved an 80-year-old math conjecture, showing AI can find novel solutions in complex problems.
• Google tops OpenAI's math breakthrough — 9 to 1 — AI systems can now autonomously generate and formally verify solutions to complex, long-unsolved mathematical problems.
• OpenAI’s math breakthrough played to AI’s strengths — AI disproved an 80-year-old math conjecture by applying broad knowledge and persistent strategy exploration, marking a significant autonomous contribution.
• OpenAI claims it solved an 80-year-old math problem — for real this time — OpenAI's new model disproved an 80-year-old math conjecture, demonstrating advanced autonomous reasoning in AI.
• An AI solution to an 80-year-old problem has shocked mathematicians — AI can autonomously solve long-standing mathematical conjectures by exploring vast solution spaces and combining existing knowledge.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Rundown AI.