What do these have in common? Algorithms.

Math that uses past data to predict the future. Sounds cool right? But there’s a problem. What happens if the data isn’t exactly pristine, or biased to begin with? Well, then the statement at the end of the computation is inevitably skewed too.

Here’s an example from a big technology company. They put together an algorithm to look through their resumes to determine the best match for employment. They discovered that the system was rejecting the female applicants. Systematical axing anyone who had attended a women’s college or had listed water polo as a sport. Anything related to a potential hireling being a woman was rejected. How did this happen? Simple, the data set was biased. Consider where you would find a group of people who could be used as a data set to teach a computer the best qualifications in the technology industry. Current employees, right? The numbers were crunched and the finding that only 14% of the technology workers are women, was immediately applied to the algorithm.