Soon after a fire engulfed Notre-Dame cathedral in Paris on Monday, news outlets began streaming live broadcasts on YouTube. Below several of the clips was an odd box of text: A snippet from Encyclopedia Britannica about the Sept. 11 terrorist attacks.
YouTube, a division of Alphabet Inc.’s Google, introduced this text box feature last year to combat the spread of conspiracy theories, including those that question the 9/11 attacks. On Monday, YouTube’s software mistakenly labeled the plumes of smoke in Paris as footage from 2001, triggering the panel below the video.
“We are deeply saddened by the ongoing fire at the Notre-Dame cathedral,” said a YouTube spokesman. “These panels are triggered algorithmically and our systems sometimes make the wrong call. We are disabling these panels for live streams related to the fire.”
Who are the humans to say the AI was wrong on this one? It found a correlation between the two topics and made its decision. The exact reasons for its decision are inscrutable to the humans, they are just contradicting the AI based on preconceived bias or because they don't like the resulting discussion.