Take a fresh look at your lifestyle.

DailyMail.com wins groveling apology from Facebook

73

DailyMail.com wins Facebook apology for AI flaw that tagged website news video of black men being harassed and arrested as ‘primates’

  • Facebook issued a public apology to DailyMail.com and MailOnline on Saturday
  • Facebook’s AI labeled a news video featuring black men as related to ‘primates’
  • Users reacted with disgust at the highly offensive label generated by Facebook
  • “This was an algorithmic error on Facebook and did not match the content of the Daily Mail’s post,” a Facebook spokesperson said in the fawning apology.










Facebook has publicly apologized to DailyMail.com and MailOnline for adding an AI-generated label of “primates” to a website’s news video featuring black men.

A Facebook spokesperson admitted on Saturday that the error was “unacceptable” and told DailyMail.com: “We apologize to anyone who has seen these offensive recommendations and to the Daily Mail for subjecting the content to them. ‘

“This was an algorithmic error on Facebook and did not match the content of the Daily Mail’s post,” the company admitted.

It concerns a news video that the Mail posted to Facebook on June 27, 2020, which depicts a white man appearing to be harassing a black man and calling the police on him.

Facebook has publicly apologized to DailyMail.com and MailOnline for adding an AI-generated label referring to “primates” to this Mail news video depicting black men

It concerns a news video posted on Facebook by DailyMail.com on June 27, depicting a white man appearing to be harassing a black man.  Facebook CEO Mark Zuckerberg can be seen above

It concerns a news video posted on Facebook by DailyMail.com on June 27, depicting a white man appearing to be harassing a black man. Facebook CEO Mark Zuckerberg can be seen above

While the video has nothing to do with monkeys or other primates, Facebook’s AI-generated prompt asked viewers if they wanted to “keep watching videos about primates,” which horrified and insulted users of the social network.

Initially, Facebook failed to notify the Mail of its blunder when it was spotted by the New York Times, and the social media giant’s statement to the Times did not definitively specify that Facebook had added the offending label to the Mail’s post.

Facebook has now disabled the faulty recommendation algorithm.

“This was clearly an unacceptable error on Facebook and we disabled the entire topic recommender feature as soon as we realized this was happening so we could investigate the root cause and prevent it from happening again,” the Facebook spokesperson told DailyMail.com.

“As we said, while we’ve made improvements to our AI, we know it’s not perfect and we need to make more progress,” the company added.

Facebook has been criticized in the past for its AI systems and alleged bias against people of color.

Studies suggest that facial recognition systems have more difficulty identifying them, leading to incidents where black people have been discriminated against or arrested because of biased AI algorithms.

The offensive label that Facebook’s AI algorithms applied to the Mail video drew strong backlash, with many criticizing the social media company for failing to monitor its automated technology.

“This really has to stop,” tweeted Charles Onyango-Obbo, journalist and former editor of Mail & Guardian Africa in Johannesburg.

“Every day we see the limits, prejudices and sheer idiocy of relying solely on AI. And yet we let ourselves be guinea pigs,” tweeted New York Times cybersecurity reporter Nicole Perlroth.

Kate Conger, a tech reporter for the New York Times, wrote, “The fact that this has happened at multiple major tech companies really says something about how AI is being built.”

“Facebook is in a state of constant resetting of their ‘how many days since we did something racist’ back to zero,” another person tweeted.

Sean C. Griffith, a lawyer, wrote that artificial intelligence “generally has a lot of problems, and tech companies often miss the race-related issues because they hire very few black people.”

“A Black #AI developer, for example, would probably have caught this one,” he noted.

Apologies from Facebook to DailyMail.com

A Facebook spokesperson shared the following statement on Saturday:

This was clearly an unacceptable error on Facebook and we disabled the entire topic recommender feature as soon as we realized this was happening so we could investigate the cause and prevent it from happening again.

This was an algorithmic error on Facebook and did not match the content of the Daily Mail’s post.

“As we said, while we’ve made improvements to our AI, we know it’s not perfect and we need to make more progress.

“We apologize to anyone who has seen these offensive recommendations and to the Daily Mail for the fact that their content is subject to this recommendation.”

Advertisement

.

Leave A Reply

Your email address will not be published.