Is Twitter’s Cropping Algorithm Racially Biased?

Recent reports are suggesting that the algorithm that controls image cropping on the Twitter app does not respond to images of dark-skinned people in the same way it does for light-skinned people. According to the TNW website, people have been noticing cropping anomalies that may have to do with the color of the featured person’s skin-tones. If you read the article, you’ll see that the jury is still out on accusation of bias, but it does raise interesting questions about our reliance on AI (Artificial Intelligence) software and how the results may reflect unconscious bias of the programmers.

While the concern about bias by Twitter’s algorithm may be unfounded, it raises additional questions about AI and facial recognition software. According to The Next Web, “Light skin bias in algorithms is well documented in fields ranging from healthcare to law enforcement.”

If you want to know more about implicit bias, see this website and take a test (I recommend the Skin-tone IAT). You may find that the best place to begin the war against bias is not Twitter’s AI software, but our own “natural intelligence.”

Leave a Reply

Your email address will not be published. Required fields are marked *

css.php