Twitter photo crop algorithm dropped due to age, weight bias

An algorithm used by Twitter to crop photos and other images on user timelines came under fire for perceived gender and racial bias.

The company's 'saliency' algorithm decided how images would be cropped in Twitter previews. However, researcher Bogdan Kulynyc discovered that the algorithm preferred to show faces that were slimmer, younger, and with lighter skin, and seemed to favour stereotypically feminine facial traits.

Kulynyc also discovered that the “saliency” of a face in an image could be increased, making it less likely to be hidden by the cropping algorithm, by making the person’s skin lighter or warmer and smoother.

Earlier, Twitter’s own research had found the algorithm had a bias towards cropping out black faces. When two faces were in the same image, the preview crop appeared to favor white faces, hiding the black faces until users clicked through. Twitter’s own subsequent analysis showed a "4 percent difference from demographic parity, in favor of white individuals".

In response, Twitter decided to scrap the image-cropping system; Twitter’s then director of software engineering, Rumman Chowdhury, stated that "how to crop an image is a decision best made by people". 

This controversy highlighted the challenges and potential pitfalls of using algorithms in social media platforms and the importance of continually testing and refining these systems to ensure they are fair and unbiased.