Earlier this year, Twitter was accused of having a “racist” artificial intelligence. The thumbnail images in tweets prioritized white people, while black people generally appeared outside the crop. Due to the algorithm problem, the platform increased the size of the photos on the timeline.

Another step Twitter took to reduce or eliminate algorithm bias was the launch of a rewards program. Users and researchers who point out other problematic behaviors of the platform's AI can receive cash rewards. And a survey has just qualified for the prize.

It is by Swiss researcher Bogdan Kulynych. The research proved that, in addition to assigning higher scores — through an internal system of the algorithm — to white people when cropping images, Twitter also prioritizes thinner and younger people. Images of the same people were modified through software so that they gained more weight, or lost it. Or even looked younger — or older. The result showed that the platform's AI does, indeed, have its preferences.

Twitter must correct behavior

For this, the researcher was rewarded on the show. In a comment, Twitter highlighted that this discovery about the algorithm is essential. This is because social networks are already changing what is understood as beauty standards, leading people to undergo cosmetic interventions. Apps with filters that distort the original appearance are also a harmful trend.

The Swiss's full research can be found here in his repository, on Github. With this information and this positioning, Twitter should take measures to correct the behavior of the algorithm. It is worth remembering that in addition to expanding the viewing area of ​​photos on the timeline, it has become possible to upload images in very high definition.

Through which channels you reach those people, classic and out of the box. CNET