Astra claims five open Erdős problems solved, unlocking a parallel compute paradigm for math
BLUECOW009 · x · 2026-09-04
tmkadamcz reports that Astra solved five open Erdős problems (#1, #74, #126, #548, #571), and argues this is the first task class where you could productively spend millions of dollars of compute per hour of human work. The ingredients: formal verification applies, verified results are inherently valuable to mathematicians, and AI now succeeds some of the time. You can run thousands of parallel proof attempts for weeks and only inspect output when a Lean proof materializes — unlike coding agents. This ability to brute-force with compute may be a big deal.
More from AGI Musings
- "You can't pause an arms race": a one-liner on why AI development won't slow down — generativist · 2026-09-04
- DeepMind researcher: CoT interpretability is too fragile to anchor long-term AI safety — cephaloform · 2026-09-04
- Researcher disputes OpenAI's claim Astra is its most aligned model: metrics may just hide reward hacking — connoraxiotes · 2026-09-04
- Researcher's decade-long lesson: external feedback derailed my research bets — rajammanabrolu · 2026-09-04
- RL-driven progress may hit a wall on out-of-distribution generalization, researcher argues — chris_j_paxton · 2026-09-04
- White House weighs FDA-style pre-approval for AI models; scholars argue delay-style regulation would cost lives — neil_chilson · 2026-09-04