A beauty contest that was meant to be judged on aspects such as on facial symmetry, gender, age and smoothness of skin didn’t go quite as planned when the judges – by that we mean robots – selected winners who had white skin.

Beauty.AI was a tool created to see what human beauty looked like from the perspective of artificial intelligence.

After 6,000 people from more than 100 countries submitted photos for the contest, it appears that the algorithms seemed to prefer white people as the benchmarks for human beauty.

Of the 44 winners, only a few of them were Asian and only one had dark skin.

Beauty AI.
(Beauty AI/YouTube screenshot)

The results have sparked controversy as to whether programmers are writing algorithms that may be unconsciously biased.

Beauty.AI was created by Youth Laboratories and supported by Microsoft. They amassed a large dataset of photos and used it to build complex algorithms to assess human attractiveness.

The judges, consisting of a five-robot panel, then picked winners from the photos submitted by volunteers using algorithms to analyse specific traits that they perceived were relevant to outer beauty.

So the question is, how did the robots select white people as their winners?

Beauty AI.
(Beauty AI screenshot)

According to Konstantin Kiselev, chief technology officer of Youth Laboratories, it could have been because the data used to establish standards of beauty did not include enough minorities.

“We had this problem with our database for wrinkle estimation, for example,” Kiselev told Motherboard.

“Our database had a lot more white people than, say, Indian people. Because of that, it’s possible that our algorithm was biased.”

Kiselev added that 75% of the entrants were European, while 7% were from India, and only 1% were from the African continent.

Meanwhile, Alex Zhavoronkov, Beauty.AI’s chief science officer, told The Guardian he was as surprised with the results as anybody else.

“When you’re training an algorithm to recognise certain patterns… you might not have enough data, or the data might be biased,” he said.

“I was more surprised about how the algorithm chose the most beautiful people.

“Out of a very large number, they chose people who I may not have selected myself.”