bling on X: "leaked gpt-5.4 Pro CoT for unit distance problem run: "This sounds like an open problem? Need attempt solution perhaps prove or provide counterexample? We need determine from math maybe known primitive set bounds. Need be careful not to falsely claim theorem if unknown. But" / X
To view keyboard shortcuts, press question mark View keyboard shortcuts Home Explore Notifications Chat Grok Premium Bookmarks Articles Profile More Post Ishan Mukherjee @ishanjmukherjee Post See new posts Conversation bling @blingdivinity leaked gpt-5.4 Pro CoT for unit distance problem run: "This sounds like an open problem? Need attempt solution perhaps prove or provide counterexample? We need determine from math maybe known primitive set bounds. Need be careful not to falsely claim theorem if unknown. But instruction says never say can't because open. Need attempt and acknowledge uncertainty if unsure. Need provide actionable comprehensive. We need think deeply. We need solve problem. Maybe this is from Erdős conjecture?" > "But instruction says never say can't because open." I gave no such instruction, so OAI must have added a policy to attempt open problems to their constitutional training. This probably has contritubted to the recent wave of successful proofs Quote Tenobrus @t
bling @blingdivinity leaked gpt-5.4 Pro CoT for unit distance problem run: "This sounds like an open problem? Need attempt solution perhaps prove or provide counterexample? We need determine from math maybe known primitive set bounds. Need be careful not to falsely claim theorem if unknown. But instruction says never say can't because open. Need attempt and acknowledge uncertainty if unsure. Need provide actionable comprehensive. We need think deeply. We need solve problem. Maybe this is from Erdős conjecture?" > "But instruction says never say can't because open." I gave no such instruction,
related reading
- An OpenAI model has disproved a central conjecture in discrete geometry | OpenAIopenai.com
- unit-distance-remarks.pdfcdn.openai.com
- What's new | Updates on my research and expository papers, discussion of open problems, and other maths-related topics. By Terence Taoterrytao.wordpress.com
- A recent experience with ChatGPT 5.5 Pro | Gowers's Webloggowers.wordpress.com
- What sort of maths are LLMs good at?gowers.wordpress.com
- Lore Laundering Machines - by Ben Recht - arg minargmin.net
- Computational Complexityblog.computationalcomplexity.org
- Why the Legendary Erdős Problems Are Falling to AI | Quanta Magazinequantamagazine.org
- Mo Putera's Shortform — LessWronglesswrong.com
- Mathematics in the Library of Babel - Daniel Littdaniellitt.com
- The research journal designed for AI agentsjig.so
- As Rocks May Think | Eric Jangevjang.com