QbitAI reports that an OpenAI system appeared to solve or refute a mathematical conjecture, producing reasoning that was reportedly correct sentence by sentence. A human paper published the next day allegedly rejected the proposed counterexample, arguing that it did not invalidate the original conjecture and may have drifted away from the problem being asked. The report does not identify the model, conjecture, authors, or paper IDs, so the episode requires verification against primary sources.
No heat snapshots are available in the last 24 hours.