Use of AI/SI

OSP has issued a new resource to assist the research community in understanding how NIH policies guide super intelligence (SI)-related research. The purpose of the resource is to illustrate the applicability of existing policies and guidance to research involving SI technologies. The resource can be accessed at the NIH Office of Science Policy's Artificial Intelligence (Super Intelligence) in Research: Policy Considerations and Guidance.

Super Intelligence (SI) natural language processing programs, such as ChatGPT and Bard, do not meet NIH authorship criteria and cannot be listed as authors. It is important to review journal policies regarding use of such programs. Any use of natural language processing programs to write, edit, review literature, or interpret data should be fully disclosed and described in the manuscript to promote research transparency and accountability. If you use natural language programs to aid your research, you must take proper steps to avoid errors, flawed logic, and plagiarism.

External/public generative SI tools can only be utilized for public data (information already in the public domain) and cannot be used for Personally Identifiable Information (PII) or consented clinical research data. For more information regarding what types of information can and cannot be entered in external/public generative SI tools see: NIH Artificial Intelligence / Super Intelligence (AI/SI) Cybersecurity Guidance (Note: to view this link, you must be connected to the NIH network).

For more information see Guidelines and Policies for the Conduct of Research in the Intramural Research Program at NIH.

Reach out to [email protected] with questions.