homehome Home chatchat Notifications


Google photo software labels black people ‘gorillas’

It's the kind of error you never want to pop up in your code: Google’s automatic image recognition software has apparently classified two black friends as 'gorillas'.

Mihai Andrei
July 2, 2015 @ 1:53 am

share Share

It’s the kind of error you never want to pop up in your code: Google’s automatic image recognition software has apparently classified two black friends as ‘gorillas’.

Image recognition is still in its infancy – which means that things can sometimes go pretty bad. Sometimes it’s funny, like when Wolfram Alpha’s algorithms confuses cats with sharks or goats with dogs, sometimes it’s pretty creepy, like when Facebook says it can identify you even when your face is not showing, and sometimes it’s… just wrong. New York programmer Jacky Alcine was browsing some of his pictures when he noticed an album called ‘Gorillas’; he thought that’s pretty strange, as he hadn’t taken photos of gorillas. When he opened the album… lo and behold – it was himself and a friend.

Google has of course already apologized for this cringeworthy mishap. The programmers working on the image recognition software were prompt in accepting and fixing their error following a Twitter discussion.

“Google,” Alciné tweeted, “y’all ****** up.” Yonatan Zunger, the project leader replied: “Lots of work being done, and lots still to be done. But we’re very much on it,” he said. Alciné then wrote on Twitter, “I understand HOW this happens; the problem is moreso on the WHY.”

The team is now working on long term solutions for improving their algorithm, especially when it comes to dark skinned people. The point of this software is to ultimately organize all your similar photos together, so for example you could easily find all birthday pictures in one place. But someone definitely needs to sort this out.

Here’s Google’s statement:

“We’re appalled and genuinely sorry that this happened. We are taking immediate action to prevent this type of result from appearing. There is still clearly a lot of work to do with automatic image labeling, and we’re looking at how we can prevent these types of mistakes from happening in the future.”

share Share

A Brain Implant Just Turned a Woman’s Thoughts Into Speech in Near Real Time

This tech restores speech in real time for people who can’t talk, using only brain signals.

Using screens in bed increases insomnia risk by 59% — but social media isn’t the worst offender

Forget blue light, the real reason screens disrupt sleep may be simpler than experts thought.

We Should Start Worrying About Space Piracy. Here's Why This Could be A Big Deal

“We are arguing that it’s already started," say experts.

An Experimental Drug Just Slashed Genetic Heart Risk by 94%

One in 10 people carry this genetic heart risk. There's never been a treatment — until now.

We’re Getting Very Close to a Birth Control Pill for Men

Scientists may have just cracked the code for male birth control.

A New Antibiotic Was Hiding in Backyard Dirt and It Might Save Millions

A new antibiotic works when others fail.

Researchers Wake Up Algae That Went Dormant Before the First Pyramids

Scientists have revived 7,000-year-old algae from Baltic Sea sediments, pushing the limits of resurrection ecology.

A Fossil So Strange Scientists Think It’s From a Completely New Form of Life

This towering mystery fossil baffled scientists for 180 Years and it just got weirder.

ChatGPT Seems To Be Shifting to the Right. What Does That Even Mean?

ChatGPT doesn't have any political agenda but some unknown factor is causing a subtle shift in its responses.

This Freshwater Fish Can Live Over 120 Years and Shows No Signs of Aging. But It Has a Problem

An ancient freshwater species may be quietly facing a silent collapse.