Microsoft Warns Washington to Regulate A.I. Before It's Too Late

Last year, at the University of St. Thomas in Minneapolis, Minnesota, a graduate student set up an artificial intelligence system based on Microsoft's facial-recognition tools, which he used to predict their emotional state. The idea, the university explained at a national education conference in February 2018, would be to allow teachers to gather real-time information on how their lessons were being received, in order to maximize "student engagement." Although the program was only tested as a short experiment, and was never adopted by the university, such a system has the potential to be rife with flaws and primed for abuse, warn experts at the research group AI Now. The St. Thomas proof-of-concept system is just one of a number of data points that AI Now--a group composed of tech employees from companies including Microsoft and Google, and affiliated with New York University--says exemplify the need for stricter regulation of artificial intelligence. The group's report, published Thursday, underscores the inherent dangers in using A.I. to do things like amplify surveillance in fields including finance and policing, and argues that accountability and oversight are necessities where this type of nascent technology is concerned.