Solving Artificial Intelligence’s Racism Problem

Several prominent voices in the world of science and technology such as Stephen Hawking and Elon Musk have warned us repeatedly about the dangers in creating machines that can grow to become more intelligent than us. But while so many are preoccupied with the future threat of artificial intelligence, few are aware of the current dangers that come with artificial intelligence: dangers that stem from artificial intelligence’s inherent racism.

Last year, users of Google’s photo app noticed that the software was automatically applying the label “gorilla” to black people. Similar errors occurred with Nikon when it’s software assumed that Asians were blinking even when they were not. And Hewlett-Packard’s web camera software has had difficulty recognizing people that have darker skin tones. These problems are relatively minor compared to the problems that can occur when artificial intelligence is used for predictive policing.

Widely used software that determines an inmate’s likelihood of relapsing into criminal activity was found to be biased against black people. It was more than twice as likely to flag black inmates as high risk. It was also twice as likely to flag white defendants as low risk incorrectly.

Many police departments already rely on artificial intelligence to try and predict the areas where crime is most likely to occur so that they can have more officers assigned to those areas. This leads to the over-policing of areas that are predominantly non-white.

The root of the problem

So far, artificial intelligence programmers haven’t been able to develop artificial intelligence that learns like humans do, so they’ve had to rely on feeding input into artificial intelligence systems. The problem is that most of this input is put together by a field that is predominantly white. As the artificial intelligence industry gathers more researchers of different ethnicity, it’s racism problem will most likely disappear.

Artificial Intelligence News brought to you by artificialbrilliance.com
Source: nytimes.com/2016/06/26/opinion/sunday/artificial-intelligences-white-guy-problem.html