OpenAI publishes 722 math papers online to address crediting concerns
OpenAI just published 722 mathematical research papers online, hoping to clear the air after some mathematicians called them out for not crediting earlier human research, especially on tough problems like the Erdos unit-distance conjecture.
By publishing their work, OpenAI wants to show how AI can actually help push math forward.
OpenAI posts protocols on GitHub
To help share the work responsibly, OpenAI teamed up with experts from the Institute for Advanced Study and used a special proof-checking code called Lean.
All 722 research papers (grouped into 372 sets) are now up on GitHub, along with protocols explaining how papers can be revised and how the results should be cited.
It's a big move meant to help mathematicians and maybe inspire more trust in what AI can do for science.