- A Fields Medalist, Timothy Gowers, tested ChatGPT 5.5 Pro by challenging it with open problems in number theory, and the model managed to improve an exponential bound to a polynomial one within under an hour.
- The key idea behind this improvement was deemed “completely original” by an MIT researcher involved in the project.
- This development marks a significant shift as Gowers highlighted that the bar for mathematical contributions is now being set at proving something that Large Language Models (LLMs) cannot do, suggesting LLMs have reached their limits in certain complex problem-solving areas.
Originally published at the-decoder.com. Curated by AI Maestro.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.