At the Institute for Advanced Study a consultative group on AI was created

In the Institute for Advanced Study (IAS) an advisory group on mathematics and artificial intelligence has started working. The organization will be engaged in coordinating the publication of significant scientific results obtained during the development of large language models.

The group's first task will be to collect proposals from the scientific community on the regulation of data disclosure related to internal research at OpenAI. Experts plan to establish standards for the "responsible presentation" of mathematical proofs that underlie the architecture of modern neural networks.

The group is focused on ensuring transparency in areas where mathematical breakthroughs can directly impact the safety and predictability of models. Participants in the initiative emphasize the need for independent verification of complex algorithmic solutions before their implementation in commercial products.

The advisory board consists of leading mathematicians and machine learning specialists. The main goal is to prevent risks associated with the publication of incomplete or unverified data, which can lead to errors in the architecture of future systems.

The creation of such a body reflects the growing demand for academic oversight in the AI industry. Previously, such processes took place in a closed mode, but now developers are striving to formalize interaction with the scientific community to increase trust in the results of their research.