I’ve been meaning to write more meaningful commentary on this, but I haven’t gotten around to it, so let me just highlight these for now.

  • Breaking the Black Box is an incredible 4-part series by ProPublica that explores hidden biases in algorithms. The themes they explore are (1) Facebook’s data collection, (2) prices varying according to zip code or operating system, (3) machine learning by A/B testing, and (4) AI systems that learn to be racist based on the inherent bias in the data given to them. The whole series is worth reading; they’re short and have little interactive toy examples at the end of each one.

  • What Happened When Dylann Roof Asked Google For Information About Race? Basically a case study of (3) above, but with very real-life consequences. In particular, Roof writes the following in his ‘manifesto’:

    But more importantly [the Trayvon Martin case] prompted me to type in the words “black on White crime” into Google, and I have never been the same since that day. The first website I came to was the Council of Conservative Citizens. There were pages upon pages of these brutal black on White murders.

    The NPR article goes on to examine how Google autocompletes “black on white” versus “white on white”, and other searches.