Reveals critical limitations in AI's ability to formalize mathematical proofs, impacting researchers relying on automated verification.
Mathematicians found OpenAI's AI-generated Navier-Stokes proof contains a mistranslation between natural language and formal Lean code, raising concerns about AI proof reliability.
Proposes creating an 'Alignment Forum for AIs' to address verification challenges and AI contributions to alignment in an era of proliferating AI-generated content.
Randomized controlled trial shows GPT-4o access improves student learning outcomes, with persistent gains when used as a tutor but fading benefits when used for automation.
OpenAI released 722 mathematical proofs generated by an internal frontier model, including solutions to 90 top open problems from Proof of Atlas.
Researchers release empirical paper testing distillation techniques for AI safety, including Distillation for Incrimination to detect hidden misalignment in models.
Explores quantum mechanics' measurement problem and wavefunction collapse as fundamental aspects of the universe's information-propagating dynamics.
