THE BLOG

Podcast: Are Robots As Racist As We Are?

07/11/2016 11:55 AM IST | Updated 20/11/2016 9:18 AM IST
NEW! HIGHLIGHT AND SHARE
Highlight text to share via Facebook and Twitter

Can you imagine a beauty contest in which the participants did not put on any make up, did not have a swimsuit round and did not give answers to the judges' questions? Well, then you probably haven't heard about Beauty.AI, which was a beauty contest judged by robots.

Out of the 44 winners, 36 were white and only a handful were Asian; only one of the winners had dark skin.

Yes, robots are now judging people according to pre-determined parameters — an idea that would have seemed far-fetched not so long ago. All the "contestants" had to do was upload selfies with conditions like they should not have any makeup on, there should be no facial hair and no glasses. After this the baton was handed over to the robots or, more accurately, Artificial Intelligence (AI).


Shutterstock


Artificial Intelligence is basically the intelligence exhibited by a machine. That is the ability of the machines to respond to humans. An good example such AI would be the chess games that we play online against a virtual opponent. Colloquially, the term "artificial intelligence" is applied when a machine does those functions that humans associate with other human minds, such as "learning" and "problem solving."

So, back to our beauty contest that was judged by AI. The contest was organised by a company from Russia called Youth Laboratories. They designed an AI system that was fed with algorithms to enable judging the contest. There were separate algorithms designed to look at wrinkles, facial symmetry, the amount of pimples and blemishes, race and perceived age. The system was fed photos of celebrities with supposedly beautiful and perfect faces to set a bar and then the contestants were judged according to that. To understand how the system worked, we need to understand the way the AI was built. For example, one of the parameters for judging the contest was symmetry of face. To that end, the system was fed hundreds of images of people with symmetrical and asymmetrical faces. Over time, the algorithm learnt how to identify a symmetrical face and when the participants sent their photos, the AI could accurately judge the symmetry of their faces. The idea of the contest was to find the people who are naturally beautiful, which is why no make-up was allowed for the photos the participants sent to the system.

The contest seems fair and unbiased, right? But when the winners were announced, the robots surprised everyone with the results — out of the 44 winners, 36 were white and only a handful were Asian; only one of the winners had dark skin. Samanth Subramanian sets to find out the cause for the "racist" results.

For more episodes of The Intersection, click here.

Also on HuffPost

Robots In 2016: These Are The Robots And Drones That Will Change Our Lives

More On This Topic