Instagram is testing AI that verifies your age with a selfie scan. It’s not perfect



CNN Business

Instagram is testing new ways to verify the age of its youngest users, including using artificial intelligence that analyzes a photo and estimates the user’s age.

Instagram Meta Property said in a blog post Thursday that AI is one of three new methods it is testing to verify the age of users on the photo-sharing site. Users must use one of the options to verify their age if they change their date of birth on Instagram from under 18 to over 18. Instagram is testing these options first with its users in the United States. It already requires users to state their age when they start using the service, and uses AI in other ways to determine whether users are children or adults.

The move is part of a constant pressure to ensure that the youngest users of the photo sharing app see age-appropriate content. It comes less than a year after revelations from a Facebook whistleblower raised concerns about the platform impact on younger users† Last year, Instagram came under fire when documents were leaked by the whistleblowerFrances Haugen, showed it was aware of how the social media site can harm mental health and body image, especially in teenage girls.

The technology comes from a London-based company called Yoti† an animated video that Instagram has posted on its blog gives an idea of ​​how Yoti’s AI age estimation works: a user is asked to take a video selfie on their smartphone (Yoti said this move serves as a way to ensure that a real person in the resulting image), and Instagram shares an image of that selfie with the company. Yoti’s AI first detects that there is a face in the photo, then examines the facial features to determine the person’s age.

Julie Dawson, Yoti’s Chief Policy and Regulatory Officer, told CNN Business that the AI ​​was trained with a dataset consisting of images of people’s faces, along with the year and month that person was born. †Documentation the company released in May to explain its technology said it was trained on “millions of different facial images.”)

“When a new face appears, it does a pixel-level analysis of that face and then spits out a number — the age estimate with a confidence value,” Dawson said. Once the estimate is completed, Yoti and Instagram will delete the selfie video and the captured still image.

A young woman uses her smartphone while sitting outside a coffee shop in Jacksonville, Oregon.

Verifying a user’s age can be an annoying problem for tech companies, in part because many users may not have government-issued photo ID that can be verified.

Karl Ricanek, a professor at the University of North Carolina Wilmington and director of the school’s Face Aging Group Research Lab, thinks Yoti’s technology is a good application of AI.

“It’s worth trying to protect children,” he said.

But while such technology could be helpful for Instagram, a number of factors can make it tricky to accurately estimate age from a photo, Ricanek said, including puberty — which changes a person’s facial structure — as well as skin color and gender.

The recent documentation van Yoti indicates that his technology is, on average, slightly less accurate at estimating the ages of children between 13 and 17 and darker skin tones than children with lighter skin. According to Yoti’s data, the age estimate was, on average, 1.91 years lower for women ages 13 to 17 whose skin tones were categorized as the two darkest shades on the Fitzpatrick scale — a six-tone scale commonly used by tech companies to classify skin tones. – versus a mean error of 1.41 years for women in the same age group whose skin tones were the two lightest shades on the scale. For children ages 13 to 17, technology’s estimate of how old they are was on average 1.56 years lower, according to the document. (For teens in general, the average error rate is 1.52 years.)

What that means in practice is that there will be a lot of mistakes, said Luke Stark, an assistant professor at Western University in Ontario, Canada, who studies the ethical and social implications of AI. “We’re still assuming an average absolute error, anyway, from a year to a year and a half,” he said.

Several CNN employees — all adults over the age of 25 — tried a online demo of Yoti’s age estimation technology. The demo differs from the experience Instagram users will have in that it takes a selfie rather than a short video, and the result is an estimate of the age range rather than a specific age estimate, said Chris Field, Chief Marketing Officer of yoti.

The results varied: for some reporters, the estimated age range was right on target, but for others it was many years off. For example, it estimated that one editor was between 17 and 21 years old, when they are actually in their mid-thirties.

Stark is concerned, among other things, that the technology will contribute to so-called ‘surveillance creep’.

“It’s definitely problematic because it conditions people to assume they’re going to be monitored and assessed,” he said.

Leave a Comment

Your email address will not be published. Required fields are marked *