October 8, 2024
NIST stresses context in evaluating safety of AI systems at innovation symposium
The National Institute of Standards and Technology is stressing the importance of context and use-cases when evaluating the safety of artificial intelligence systems, at a symposium on innovation held by the agency to get feedback from industry and other researchers as the agency maps out its next steps on the technology.
Google DeepMind researcher Laura Weidinger said “it's pretty clear that we do need a sociotechnical approach” to evaluating AI systems, as part of a Sept. 25 panel discussion on...