The results have been awaited for weeks, although so far OpenAI has not provided details on what problems the model addresses or when exactly they will be released. In September, the company said its model had “solved more than 100 long-standing open problems across most areas of mathematics.” Published articles and other details also include summaries of the model’s reasoning, estimates of the calculation used, and statistics on the number of problems attempted. OpenAI claims the “average result” used the equivalent of three hours of thinking on ChatGPT Pro.
The AGMAI, or Advisory Group on Mathematics and Artificial Intelligence, released its first recommendations in late September, urging AI labs to publish mathematical results promptly and through established academic channels when possible, while disclosing details such as the name of the model used, prompts and calculation costs. The group also implored AI companies to “refrain from viewing the publication of mathematical results as a marketing means to promote their models,” a practice they say causes significant harm to the mathematics community.
Here’s how OpenAI described its process for this release:
For this release, we are publishing the results in a GitHub repository, with protocols for paper reviews and citations. We continue to explore other community-hosted alternatives for this release that meet the committee’s guidelines. For future versions, we are committed to further improving the quality of articles via citations, mathematical exposition, and presentation of results for better understanding.
Gn bussni
