This post discusses the recent the National Telecommunications and Information Administration (NTIA) AI Accountability Policy Request for Comment, relevant from an AI ethics perspective. The NTIA document, we are told, "leans heavily on two others: the Blueprint for an AI Bill of Rights from the White House Office of Science and Technology and the AI Risk Management Framework from the National Institute of Standards and Technology (NIST)." The main criticism is that the recommendations depend mostly on auditing. "For both old-school and modern AI, auditing is only one line of defense, and that's not enough. You can audit until you're blue in the face, stuff will still get through, and AI systems will still cause some harm."
Today: 0 Total: 18 [Direct link] [Share]
