Editorial illustration for Fields Medalist: ChatGPT 5.5 Pro produced PhD-level math proof in under an hour
Fields Medalist: ChatGPT 5.5 Pro produced PhD-level math...
The clock had barely ticked past an hour when Timothy Gowers received the proof. A Fields Medalist, one of the most decorated mathematicians alive, had watched a machine do what takes a PhD candidate years. No hints, no guidance, no human hand steering the logic.
ChatGPT 5.5 Pro simply opened a file, chewed on an unsolved problem in number theory, and spat out a complete scientific paper. Gowers didn't touch a single equation. The AI worked alone, and it worked fast.
This isn't a prediction. It's a published result.
According to Gowers, ChatGPT 5.5 Pro thought for 17 minutes and 5 seconds, then delivered the best possible construction with a quadratic bound.
And so the question ceases to be whether machines can *assist* with math. Gowers’ experiment answers that: they can, for certain problems, do the work themselves. The deeper implication is not about job displacement, it’s about speed.
What happens when a field that measures progress in years suddenly gains a tool that collapses centuries of literature into a single afternoon? We are not looking at a faster calculator. We are watching the very definition of mathematical creativity being rewritten, line by line, in real time.
Related Reading
Browse all OpenAI news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic OpenAI hires 630 ex-Meta staff as ChatGPT memory may turn data into ads
Business & Startups
Related Topic Sen. Warren urges Trump admin to expand AI credit after OpenAI letter
Business & Startups
From Archives Key Topics for LLM Engineers: Using Instruction Data to Align Models
From Archives Semantic memory query retrieves Friday deployment approval for user-123
From Archives OpenAI launches Realtime‑Translate for 70+ languages and Realtime‑Whisper transcription
From Archives Anthropic, OpenAI launch joint venture as enterprise AI rush intensifies
Business & Startups
Common Questions Answered
How long did ChatGPT 5.5 Pro take to produce a PhD-level math proof according to Timothy Gowers' experiment?
ChatGPT 5.5 Pro completed a PhD-level mathematical proof in under an hour, working entirely independently without any human guidance or hints. This represents a dramatic acceleration compared to the years typically required for PhD candidates to develop such proofs.
What type of mathematical problem did ChatGPT 5.5 Pro solve in Timothy Gowers' experiment?
ChatGPT 5.5 Pro tackled an unsolved problem in number theory and produced a complete scientific paper as its solution. The AI worked autonomously on this complex mathematical challenge without any human intervention or steering of the logic.
What is the broader implication of ChatGPT 5.5 Pro's mathematical capabilities beyond job displacement?
According to Gowers, the deeper implication is about speed and how mathematical progress is measured. When a field that traditionally measures advancement in years suddenly has a tool that can synthesize centuries of literature and solve problems in an afternoon, it fundamentally rewrites the definition of mathematical creativity and research methodology.
How does ChatGPT 5.5 Pro's mathematical ability differ from being a faster calculator?
ChatGPT 5.5 Pro is not simply a faster computational tool but rather demonstrates genuine mathematical problem-solving capability that can independently work through unsolved problems and produce complete scientific papers. This represents a qualitative difference in AI capability, showing that machines can now perform the creative mathematical work itself rather than just accelerating calculations.
Further Reading
- A recent experience with ChatGPT 5.5 Pro — Gowers's Weblog
- AI Has Reached Research Mathematics. Not as a Genius. As ... — abvx.substack.com
- GPT-5.4 Solved a 60-Year Math Problem: What Happened — Build Fast With AI
