Identity theft's newest target: your face

A lot of companies struggle with bias in the workplace, but for many big tech companies, the problem is a bit more extreme. Why, because it's not just the human beings that are racist, a lot of their algorithms are biased too. This is the problem that Google reportedly was trying to solve when they got themselves into their latest privacy scandal, tricking black and brown volunteers to submit to 3D face scans.

According to the New York Daily News, Google contractors told these biometric test subjects that they were just going to play a selfie game, and they'd even get paid for it. Paid all of five bucks so that Google could use their identity to build more accurate facial recognition, holding onto their facial model forever. And it gets even worse, not only were contractors reportedly lying, but they also targeted the homeless and others in dire financial straits, hoping they were less likely to complain publicly.

The whole reason Google did this to begin with is because facial recognition is more than a little bit biased. We think of software as objective, but the truth is that artificial intelligence tools like facial recognition depend on thousands of subjective choices made by human beings who build them. Those human choices can automate discrimination, as we see facial recognition products on the market that are dozens or even hundreds of times less accurate for women, trans, people of color, children, and other groups. In short, these tools often only work well for middle-aged white guys. Read the rest