NSF Reveals New Guidelines on Generative AI Use in Merit Review for Research Proposals

The National Science Foundation has issued a notice informing the research community of new generative artificial intelligence guidelines for reviewers and proposers to protect the integrity of proposals and their evaluation in the NSF merit review process. 

The new review procedure disallows NSF reviewers from uploading any proposal content, review records and related information to non-approved generative AI solutions. The NSF memo issued on Thursday also encourages project proponents to indicate in their descriptions the extent of generative AI use in developing their proposals, FedScoop reported.

An NSF spokesperson told FedScoop that the foundation’s approved generative AI tools are publicly accessible solutions explicitly for public information use. 

The NSF noted that it is still looking into ways to integrate generative AI applications in its business processes and determining whether a commercial or an in-house solution is the answer.

In October, the foundation announced the completion of a study assessing data and analytics applications to determine its future investments. NSF Director Sethuraman Panchanathan said advanced data and analytics’ integration into decision-making and assessment could transform how the foundation invests in research and innovation projects.