The problem occurred in June 2020, when Croatian chess player Antonio Radic’s YouTube channel was blocked while presenting a video with discipline teacher Hikaru Nakamura, a decision that was reversed 24 hours later. The YouTube platform algorithm blocked chess videos by understanding that “white” and “black” were discriminatory terms. The misunderstanding was not the product of human failure, but of a response from the artificial intelligence system that moderates the content in the video portal, which presents a mixed system of control between humans and algorithms. The information was recently released following research from Carnegie Mellon University, Pennsylvania, USA, when specialists discovered that the platform’s artificial intelligence system reviewed the conversations in the videos and found phrases such as “white versus black”, and could not tell that it was the color of the chips at stake. By recording that exchange, YouTube categorized the content as a hate speech. The conclusions of the study were presented at the Conference “Association for the Advancement of AI”. After manually reviewing a selection of 1000 comments that YouTube’s artificial intelligence had classified as hate speeches, they found that 82% of them had been misclassified due to the use of words such as “black”, “white”, “attack” and “threat”, usual terminology in the board game. In defense of artificial intelligence, one of the researchers at the Institute of Language Technologies, who participated in the conference explained that such errors can happen, and also happens on other networks such as Facebook or Instagram where they daily censor posts erroneously.
translated from Spanish: Unusual Youtube error: blocked chess games for recording discrimination
February 27, 2021 |