NIST Faces Major Hurdles in Quest to Standardize Trustworthy AI Practices
-
NIST is a small federal agency tasked with creating standards for safe, secure, and trustworthy AI. This is a major challenge.
-
NIST published an AI Risk Management Framework last year that catalogued risks like bias and privacy threats.
-
NIST chief AI advisor Elham Tabassi stresses the need for a shared lexicon and input from various disciplines to create successful AI standards.
-
NIST has over a dozen people working on AI standards and is expanding, but faces tight deadlines and budget constraints.
-
NIST aims to help industry develop standards, not regulate AI, but will conduct some model evaluations. Safeguards must be built in early in development.