Machine learning algorithms can be sexist

Machine learning is the process of feeding data into a program so that it ‘learns’ how to perform a certain task, without engineers having to explicitly program an algorithm. A few years ago, companies like Google and Microsoft open-sourced their machine learning platforms, allowing anyone to apply machine learning tools to their own projects. This year, we’ve really started to see the fruits of those accessible platforms, with machine learning applied to a wider range of tasks that ever before.

Machine learning is now being used, for instance, to detect the sounds of illegal logging in the rainforest, monitor the health of dairy cows, and detect cancer cells that are too small for pathologists to see.

However, machine learning isn’t without its share of controversies. It recently emerged that Amazon had to scrap an algorithm it was using to sort job applications, because it became biased towards male candidates. The algorithm was trained using employment data from the past 10 years at Amazon, and thus reflected male dominance in the tech industry. Machine learning is a powerful tool, but it’s only as good as the data on which it’s trained.