📊 poll results
AI just disproved an 80-year-old Erdős math conjecture, verified by a Fields Medalist — but OpenAI's last "Erdős breakthrough" claim was debunked. Real progress or hype cycle?
🗳️ 19 votes so far 🏆 Leading: This is real, verified original discovery. A genuine sign AI can now generate new science, not just recombine what's already known — and the human verification process by Fields Medalist Tim Gowers and others is exactly how it should work. ⚖️ Split decision

Of 19 votes cast so far, This is real, verified original discovery. A genuine sign AI can now generate new science, not just recombine what's already known — and the human verification process by Fields Medalist Tim Gowers and others is exactly how it should work. leads with 42%.

Handle with skepticism regardless of verification. OpenAI's own track record here includes a debunked claim just seven months earlier, and headline math "breakthroughs" tend to get overstated in the announcement before the dust actually settles. is second on 37%, close enough that a handful of votes could still turn it around.

Everyone who has answered so far is voting from China.

The most recent vote came in 7 days ago.

hewsay who say wut? 💬
📊 poll results
AI just disproved an 80-year-old Erdős math conjecture, verified by a Fields Medalist — but OpenAI's last "Erdős breakthrough" claim was debunked. Real progress or hype cycle?
42%
Leading answer
This is real, verified original discovery. A genuine sign AI can now generate new science, not just recombine what's already known — and the human verification process by Fields Medalist Tim Gowers and others is exactly how it should work.

Discussion

react to this tea 👀
This is real, verified original discovery. A genuine sign AI can now generate new science, not just recombine what's already known — and the human verification process by Fields Medalist Tim Gowers and others is exactly how it should work.
Math is a uniquely easy domain for this. It's the one field where a computer can formally verify an answer is airtight, like DeepMind's Lean-checked proofs the very next day. That doesn't mean AI can do open-ended science in fields where there's no ground truth to check its work against.
Handle with skepticism regardless of verification. OpenAI's own track record here includes a debunked claim just seven months earlier, and headline math "breakthroughs" tend to get overstated in the announcement before the dust actually settles.