Report by James Lewis: “…Criticism of FRT is too often based on a misunderstanding about the technology. A good starting point to change this is to clarify the distinction between FRT and facial characterization. FRT compares two images and asks how likely it is that one image is the same as the other. The best FRT is more accurate than humans at matching images. In contrast, “facial analysis” or “facial characterization” examines an image and then tries to characterize it by gender, age, or race. Much of the critique of FRT is actually about facial characterization. Claims about FRT inaccuracy are either out of date or mistakenly talking about facial characterization. Of course, accuracy depends on how FRT is used. When picture quality is poor, accuracy is lower but often still better than the average human. A 2021 report by the National Institute of Standards and Technology (NIST) found that accuracy had improved dramatically and that more accurate systems were less likely to make errors based on race or gender. This confusion hampers the development of effective rules.
Some want to ban FRT, but it will continue to be developed and deployed because of the convenience for consumers and the benefits to public safety. Continued progress in sensors and artificial intelligence (AI) will increase availability and performance of the technologies used for facial recognition. Stopping the development of FRT would require stopping the development of AI, and that is neither possible nor in the national interest. This report provides a list of guardrails to guide the development of law and regulation for civilian use….(More)”.
