Pediatrician says AI slow-motion videos for infants are “garbage”

With its bright visuals and smoothly animated babies and colorful trains, you might think you’ve found a mediocre online video teaching kids their ABCs.

But this may actually be an example of meaningless content mass-generated by artificial intelligence, known as AI slop, circulating on YouTube, alarming child development experts and advocacy groups like Fair Play. This month, the group sent a letter to Google, YouTube’s parent company, expressing concerns about the decline in artificial intelligence. (new window)demands changes to the way AI videos are displayed and distributed on the platform.

AI slop It has a negative impact on a child’s development by distorting their sense of reality, overwhelming their learning process, and hijacking their attention. The letter says:

What is baby slop?

Italian people’s brains are rotten (new window) Video and other AI “slop content” (new window) While it has previously appealed to older viewers, experts say Baby Slop is the latest wave of AI-generated videos aimed directly at toddlers and preschoolers.

Kelly Ewart, an associate professor of education at Wilfrid Laurier University who studies AI literacy, said videos can include bright colors, repetitive sounds and familiar characters from popular movies and TV shows.

But she says something like that In reality, there is no developmental or cognitive coherence or intent.

Watch | How to spot the “Baby Slop” video:

How do we know it’s AI?

If you look closely at a video, you can often find classic signs that suggest it was generated by AI.

Characters and objects may suddenly become different (a red car looks different in each scene) or behave strangely (a child on a scooter disappears into the ground, a stream of underwater bubbles flows downward instead of upward). The actual letters and numbers may be mixed with meaningless characters.

When you expand the description of a YouTube video, you may see that the video contains: Modified or synthetic content — but not all creators disclose this.

How is this harmful?

Because infants and preschoolers’ brains are still in the early stages of understanding the world and developing language skills, Ewart suggested that some AI videos are so confusing that children may not be able to understand them. No meaning can be derived from it.

According to children’s content creator Carla Engelbrecht, these types of videos can model risky behavior. For example, someone who grabs a hot pot with their bare hands.

I’ve seen content that models behavior that I don’t want my children to experience. A creator living in California who has worked with said this. sesame streetNetflix and PBS Kids for 25 years.

Dr. Michelle Ponti, Canadian Pediatric Society Digital Health Task Force Chair (new window) Lead author of Screen Guidelines for Children Under 5 (new window)I’m honest when describing these videos.

It’s garbage and young children can’t learn from garbage.said the London, Ont.-based pediatrician.

We know what promotes early learning. It’s direct contact with a loving caregiver, a back-and-forth interaction with a real human being who can make a connection and help a child learn. she said. AI slop is just a collection of garbage.

Watch | How screen time affects young brains:

Ponti recently co-signed a Fair Play letter asking Google and YouTube to make changes, including clearly labeling all AI-generated content and banning AI-generated content on YouTube Kids.

YouTube CEO Neil Mohan said in January that “low quality and repetitive” AI content was attracting the company’s attention. (new window).

In a statement to CBC News this week, a YouTube spokesperson said the company high standards For YouTube Kids content and limits AI-generated content within the app A small set of high quality channels.

A spokesperson said creators on the main platform will be required to publish AI content that looks real and indicate whether they have used YouTube’s AI tools. The spokesperson also said that a labeling strategy for YouTube Kids is in the works.

CBC News attempted to contact multiple YouTube channels that create videos for preschoolers that appear to be AI-generated, but did not receive a response in time for publication.

Are there any guidelines for children’s videos?

With quality children’s programming like mr dress up, Mister Rogers’ Neighborhood and sesame street Ewart says if language has a purpose repeat with intention provide An opportunity for children to develop their understanding.

expand image (new window)

According to Engelbrecht, traditional television and online programs that incorporate child development typically follow principles of multimedia, design, and storytelling.

For example, to support literacy skills, spell out a word and then show its objects and say them on the screen to emphasize concepts for young brains. It shouldn’t be too fast-paced or too busy.

Notice that “The Dog is in the Garden” is a scene of a busy street with clouds and airplanes flying overhead. Engelbrecht said these additional elements can be distracting and drown out the original intent.

Can’t parents forbid it?

Calgary parent Sarah Lynby-Workland has been adjusting her family’s screen time after witnessing nearly seven-month-old Ben becoming engrossed in just a few seconds of infant sensory videos. She immediately stopped the video. She’s seen other parents show their kids online videos and have them do difficult tasks, like untangling or brushing their teeth.

AI video is so prevalent online that it may seem inevitable, Lembey-Workland says. An avalanche rushing towards us.

She points out that while YouTube Kids accounts have parental controls and parents need to know what their kids are watching, it’s impossible to scrutinize them every second or monitor them every hour of the day.

A lot has changed since we were kids. [watching] Saturday morning cartoons.

Mr Ponti agreed that this was more than a matter of parental guidance and said appropriate regulation would be in place. Putting child development at the forefront is required.

Can AI be used effectively?

Engelbrecht has been working with AI tools for many years, but recently dabbled in AI animation, which he says he uses in his work to help others improve their AI literacy. (new window)detect slope (new window) create better videos (new window).

Since AI models are already tuned to reflect copyrighted content, Engelbrecht doesn’t think adding safety guidelines to children’s content is overkill.

Direct ban on AI content — I think the toothpaste definitely ran out of the tube. she said. But it is possible to talk about and enforce standards and quality of content.

The current challenge, she says, is that platforms and algorithms emphasize fast-moving video and visual transitions, which is inconsistent with best practices in video production and child development.

expand image (new window)

But when pushed to incorporate such principles into generative AI, she says: It would actually do a pretty decent job.

She believes AI can be used to create online children’s content that is a middle ground between fast food and Michelin-starred cuisine.

She also mentioned the studio behind the hugely popular online video series cocomelonIncorporate child development research in collaboration with UCLA scholars (new window) It can also affect your content in the following ways: It has a huge impact and influence.