Twitter cropping algorithm shows bias towards light-skinned and younger faces.

August 11, 2021 | 1 minutes read

A Twitter image-cropping algorithm prefers to show faces that are slimmers, light-skinned and younger, research has found.

Bogdan Kulynyc won $3,500 (£2,530) in a Twitter-organised contest to find biases in its cropping algorithm.

Earlier this year, Twitter did research and found that the algorithm is biased towards people of color. The “saliency algorithm” decided how images would be cropped in Twitter previews before they open to full size upon clicking on them. 

If two faces are in the same image, the cropping algorithm favors the light-skinned image hiding the black face until the user clicked as found by many users. 

The pattern held true for images of former US President Barack Obama and senator Mitch McConnell and for stock images of businessmen belonging to different ethnicities.

Twitter’s own subsequent analysis showed a “4% difference from demographic parity, in favor of white individuals”.

According to the director of software engineering, Rumman Chowdhury, Twitter has concluded that “how to crop an image is a decision best made by people”.

Twitter cropping image bias

View original tweet on Twitter