YouTube is rife with AI-generated “educational” videos targeting children, and many of the lessons they’re imparting — if there’s a discernible message at all — could be harming their development.
Disturbing new reporting from The 74 and Mother Jones found numerous examples of AI-generated videos either peddling absolute nonsense or damaging lessons to their young intended audience.
In one video that’s supposed to be a nursery rhyme about cars, children ride without a seatbelt and walk in the middle of a road with moving cars behind them.
Another AI-generated sing-a-long video about the US’s 50 states shows garbled state names that don’t match up with the vocals, as kids are asked to learn about “Ribio Island,” “Conmecticut,” “Oklolodia,” and “Louggisslia.”
Carla Engelbrecht, who’s worked for children’s media brands like Sesame Street and PBS Kids, found other child-targeted AI videos showing a baby swallowing whole grapes, which is a choking hazard, or eating honey, which can kill infants. Another showed a baby eating an apple that oozed blood.
It’s not hard to see how the videos could be dangerous. Kids might pick up the idea that it’s okay to eat a dangerous snack, or wander into the road. And the videos are being cranked out at a whirlwind pace, while parents are increasingly relying on using platforms like YouTube to keep their kids entertained.
“We’re at the beginning of a monster problem, and we have to get hold of it quickly,” Kathy Hirsh-Pasek, a professor of psychology and neuroscience at Temple University, told The 74.
“This is not neutral content,” echoed Dana Suskind, a professor of surgery and pediatrics at the University of Chicago. “I think of this as toddler AI misinformation at an industrial scale. It’s very risky for the developing brain.”
Right now, it’s the exact scope is hard to pin down. But a report from video-editing platform Kapwing cited by The 74 estimated that 21 percent of YouTube’s feed is now filled with shoddy AI content. The channel behind the AI nursery song about cars has uploaded more than 10,000 videos since its first upload around seven months ago, averaging about 50 new videos per day.
A recent investigation from The New York Times into over 1,000 YouTube Shorts recommended to young children also offers some insight. After creating a fresh account and watching popular traditional children’s channels, the NYT found that nearly half of the videos the platform recommended featured AI visuals, suggesting either the algorithm heavily favors AI slop, that slop-peddlers know how to manipulate it, or that the content is astoundingly pervasive. Likely, it’s all of the above.
The NYT investigation similarly found that most of the AI videos targeting children purported to be educational, often promising to teach about animals and the letters of the alphabet, while presenting conflicting information. And every “mixed signal,” Engelbrecht warned, is delaying a child’s learning.
“Mixed signals means you are delaying them learning the cause and effect of a thing,” Engelbrecht told The 74. “If you learn that red is blue and blue is red, that’s a delay.”
“If you’re inconsistent, it takes that much longer to learn,” she added. “Every delay they have means everything else gets pushed back. That’s taking their executive function offline to go learn nonsense.”
Suskind, who is the author of the author of the upcoming book “Human Raised: Nurturing Connection, Curiosity, and Lifelong Learning in the Age of AI,” likened the cognitive effects to an even worse form of brain rot: a “brain stunt,” because the children are early in their cognitive development.
“Every experience is building a million new neural connections,” Suskind told The 74. “You will be unintentionally wiring the brain in incorrect ways.”
None of AI videos highlighted in The 74 story were found while using YouTube Kids, though the NYT reporting found many examples that did. In any case, they still targeted children, and many parents let their kids use the platform on a regular account. While a YouTube spokesperson told The 74 it has stricter “quality principles” for children-targeted content, many of these videos are clearly slipping through the cracks — cracks that are left wide by YouTube’s policy that only requires AI content be labeled if it appears “realistic,” which does not apply to the cartoonish style of the majority of these videos.
More on AI: Children’s Toys Are Shipping With Adult AI Inside Them
The post “Educational” YouTube AI Slop Encourages Kids to Play in Traffic appeared first on Futurism.
