Shows rigorous academic standards in AI research mathematics and the importance of error correction in foundational work.
OpenAI withdrew 3 mathematics research papers due to a sign error in a key argument, while revising 14 others with proof repairs and corrections.
Researchers introduce RECAST, a learned framework that formulates evidence construction as sequential decision process over retrieval and computation operations.
Researchers propose using stated-preference economics validity frameworks to evaluate LLMs when ground truth is unavailable, demonstrating the approach with six models on economic valuation surveys.
Researchers prove that clipped decentralized SGD achieves optimal convergence rates under heavy-tailed noise conditions while maintaining linear speed-up across agents.
Researchers introduce SciExam for ENSO, a benchmark where AI agents build climate models from observations, with six of twelve systems outperforming published models.
EmbodiedRSI introduces a self-evolving robot learning system that uses hypothesis-guided experimentation to improve robot skills through physical trial-and-error, achieving 77% success on RoboCasa365 benchmarks.
