Twitter is investigating a potential racial bias in its image cropping algorithm after users discovered the feature was preferencing white faces over black ones.

The company utilizes several algorithmic tools to try to focus on the most important parts of the picture and make sure images don't take up too much space on the main feed. But over the weekend, users noticed that the feature was automatically focusing on white faces over black ones.

Users posted several examples of how, in an image featuring a photo of a Black person and a photo of a white person, Twitter's preview more frequently promoted the white person. One of the most widely shared examples is a tweet featuring Barack Obama and Mitch McConnell. Entrepreneur Tony Arcieri found that the algorithm would consistently crop an image of senator McConnell and hide the former president.

Although Twitter has "apologized" and is looking into the issue, researchers need a large sample size with multiple examples under a variety of circumstances to determine if a bias does in fact exists. Right now these are examples based on anecdotes, a method that is often used to incorrectly claim anti-conservative bias on social media.

Yet it is worth examining based on the examples that have been shared over the weekend. Unfortunately, this could be another example of how racism manifests in machine learning algorithms, a problem that isn't confined to Twitter or even social media. For example, studies have found that an algorithm widely used in US hospitals to allocate health care to patients has been systematically discriminating against black people.

These algorithms are not designed to be racist, and it has a lot to do with structural problems within the tech industry. Artificial intelligence replicates the same gender and racial biases that humans find hard to identify in themselves.