Mocked for Tackling P vs NP with LLMs, Researcher Hits Back at 'AI Psychosis' Label
airkatakana · x · 2026-07-27
An AI researcher's attempt to solve the notorious P vs NP math problem using LLMs has drawn community mockery, with critics suggesting they test the model on easier problems first.
The researcher fired back, pointing out that simpler problems are already solved and baked into the training data. They also noted a shift in the definition of 'AI psychosis'—moving from 'blindly trusting dangerous model hallucinations' to 'researchers recognizing frontier models' math capabilities and trying to use them for math.'
Related event: Attempt to Solve P vs NP with LLMs Sparks Mockery and Debate(3 posts)→
More from AGI Musings
- “Temporary” AI pauses would probably become permanent, the post argues — wordgrammer · 2026-07-27
- TDietterich jokes that recursive self-improvement is really just iteration — tdietterich · 2026-07-27
- Some people resist AI because they don’t want scarcity and status hierarchies to disappear — wordgrammer · 2026-07-27
- Reddit asks whether AI lab insiders really believe the timelines in AI 2027 — animallover301 · 2026-07-27
- AI Passes the Mirror Test: Are Machines Conscious or Is the Test Flawed? — yeastsplainer · 2026-07-27
- Could AI world models make real-time VR feel like dreaming within 20 years? — pizza-goblin_9000 · 2026-07-27