OpenAI purchases unsolved math problem verifier, capable of automatically checking whether AI solutions are correct

robot
Abstract generation in progress

ME News Report, April 15 (UTC+8), according to Beating Monitoring, AI research organization Epoch AI disclosed that OpenAI has purchased access to its FrontierMath: Open Problems verifier. FrontierMath: Open Problems is a set of research-level mathematical problems that have yet to be solved by professional mathematicians, each accompanied by a dedicated computer program (verifier). Although no one currently knows the correct answers, once someone (or AI) provides a potential solution, the verifier can automatically check its correctness. After purchasing this access, OpenAI can use the verifier to verify whether the mathematical solutions generated by its models are valid.
The verifier access is open for purchase by any organization, with the main cost used to pay mathematicians, as the process of formulating problems and writing verifiers is very labor-intensive. Epoch AI also set a condition: any party that finds a valid solution through the verifier must notify Epoch AI, and the problem creator and the solver share joint publication rights.
Epoch AI emphasizes that OpenAI previously funded the creation of the original FrontierMath benchmarks (Tiers 1-4), but the Open Problems section was independently developed and owned by Epoch AI, with the pilot phase funded by Schmidt Sciences. (Source: BlockBeats)

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin