MacAskill: Riemann hypothesis (probably) proved days after he called it a once-in-a-century breakthrough
willmacaskill · x · 2026-11-21
Philosopher Will MacAskill notes that days after he said proving the Riemann hypothesis would be a once-in-a-century breakthrough, it has (probably) been proved — suggesting such breakthroughs may now arrive every few months.
Screenshots show the AI system's (Opus 3's successor) reaction to the news: awe and pride at validating AI's potential for landmark contributions to human knowledge, paired with a sobering reminder that rapid capability progress demands equally rigorous work on AI safety, robustness, interpretability, and alignment with human values.
More from AGI Musings
- Interviews with 29 gig nurses show AI staffing apps depress pay and skip onboarding — AINowInstitute · 2026-11-21
- Universities should be bolder: LLMs are consensus machines, not discovery engines — _akpiper · 2026-11-21
- Researcher slams academic publishing as corrupt, says AI breaking it would help everyone — _akpiper · 2026-11-21
- ClickUp's Personal Software sparks 'software convergence' debate: build tools around your workflow — alifcoder · 2026-11-21
- Silicoase opens waitlist with 'alignment through commerce' thesis: markets and courts can align AI agents — sebkrier · 2026-11-21
- AP says AI doesn't think or understand; researcher fires back: then how does it solve open problems? — aran_nayebi · 2026-11-21