Search and filter through extracted claims from AI researchers.
Showing 1481-1500 of 2685 claims of type "fact"
Niche construction alone does not break the complexity plateau in fitness-free evolutionary systems
k-means++ achieves O(1)-approximation with constant probability in a budget-smoothed setup
Claude Mythos preview found weaknesses in HAWK and a weaker version of AES
Many previously hyped AI models would now be recognized as not remotely godlike in hindsight
Every previous time the community went wild about an untried model, disappointment followed
Math problems are more amenable to formal verification and synthetic data than most other problems
Most problems are not as amenable to formal methods and data augmentation as math is
The cost of generating proofs for all 10 breakthroughs combined was under $2,000 at Sol API prices
The total cost of solving these ten advances was about $2000 at Sol API prices
A new tool called 'smevals' enables running small eval suites against models, harnesses, and prompts
Pipeline data may be stale or degraded.
Last synthesis: 2026-09-20. 8,985 pending.