Give the Fun Back

Go watch this week's comic first. See the four panels. A boy asks a robot to finish his jigsaw, the robot completes it instantly, and the boy stands there holding a finished picture that somebody else assembled. Read the trade once more, because a real business runs it on small children, for fractions of a cent in watch time.

The recipe

Bloomberg looked inside this trade in late 2025, and the reporters found a method simple enough to teach in an afternoon. One creator, Monique Hinton, demonstrated the entire recipe to her million followers¹. She asks ChatGPT for “simple, repetitive” children's songs with “playful nonsense words like la la, na na or clap clap”¹. The lyrics go into a video generator. Advertising revenue arrives on the other side¹. Hinton acknowledges that the machine performs roughly 95 percent of the work¹. The channel stays faceless, and an automated account publishes six or seven clips a day. One such account launched in August 2025, and by late January it had published 7,770 videos, roughly one every half hour for five consecutive months². Each view earns a fraction of a cent. The accumulated total is enormous. Fairplay, a children's advocacy organization, calculates more than $4.25 million a year flowing to the top channels built for babies³. Bloomberg estimates the wider slop economy at $117 million and 221 million subscribers¹.

On TikTok the recipe sells as an opportunity, in tutorials promising easy money to anyone with a laptop. The children's category gets recommended for one reason: repetition. A toddler replays a jingle forty times, so a single clip earns what an adult video would need forty separate viewers to produce. The objective sits in plain sight. Build the cheapest loop, aim it at the youngest audience, and hold them there. Deloitte named the risk in its 2026 predictions: "Generative video could empower independent creators and boost platforms' ad revenues, but it also risks overwhelming audiences, eroding authenticity, and fueling misinformation, likely intensifying regulatory scrutiny"⁸.

Built to hold

The videos hold attention, and that is the entire engineering objective. A New York Times investigation examined more than 1,000 shorts produced for small children. Reporter Arijeta Lajka summarized the finding: “The YouTube algorithm is pushing bizarre, often nonsensical A.I.-generated videos targeting children”². After a single CoComelon video, roughly 40 percent of the clips autoplay delivered next were machine-generated². Many displayed warped faces, extra limbs, and words that dissolved halfway through a sentence. The generators behind them cost close to zero to operate. Sora, Veo and Runway each produce a finished clip from one line of typed text⁵. MIT Technology Review has documented the same flood across every major platform this year⁵. An adult scrolls past it. A baby receives it as instruction, because toddlers learn how the world behaves by observing what the world does, and a screen of dissolving faces and broken language is teaching something specific.

What play teaches

The comic strip makes the argument better than a chart can. A child solving a puzzle learns patience, searches for the edge piece, rotates it three ways, and feels the click when it finally seats. Solve the puzzle for her and the lesson departs with the fun. Development runs entirely on that effort. Common Sense Media counted the hours in 2025, and found children under eight watching about two and a half hours of screens a day, with four in ten two-year-olds holding their own tablet⁶. JAMA Pediatrics followed 7,100 one-year-olds, and found more screen time at age one tracking with speech and problem-solving delays at ages two and four⁷. Put the two findings together. The hours the feed consumes are the same hours a developing brain would otherwise spend learning to reason. Watch time climbs. The capability a puzzle builds recedes.

Pushback

Pressure is accumulating on YouTube. Neal Mohan, the chief executive, identified AI slop as a priority in his January 2026 letter⁴. A policy from July 2025 now classifies mass-produced, repetitive uploads as “inauthentic content,” and the platform used it to remove sixteen channels at once, accounts holding roughly 35 million subscribers and 4.7 billion views between them⁴. YouTube is also testing a prompt that asks viewers to rate how much a clip qualifies as slop⁴. Fairplay assembled more than 200 organizations behind a letter to Mohan and Sundar Pichai, requesting firmer protections around the youngest users³. Deloitte expects Washington to act next, through age verification requirements, renewed argument over the 1996 liability shield, and mandatory labels on AI video⁸.

By the numbers

• 40 percent of the shorts served to small children after a nursery hit were machine-generated².
• $4.25 million a year flows to the top AI channels made for children³.
• 1,000 shorts fed to toddlers formed the sample for that count².
• Two and a half hours of screens a day is the average for kids under eight⁶.
• 40 percent of two-year-olds own a tablet today⁶.
• 7,100 one-year-olds in a JAMA study showed more screens tracking with delay⁷.
• 16 channels, 35 million subscribers, and 4.7 billion views fell in one YouTube purge⁴.
• 90 percent of web content could be machine-generated by 2026, on one Europol estimate⁹.

What comes next

The trend points one direction. Europol has estimated that up to 90 percent of web content could be machine-generated by 2026, and the generators keep growing cheaper and faster⁹. Deloitte expects fresh laws on age checks and content labels⁸. Call 2027 and 2028 the sorting years. In the better version, platforms label genuine human work, a subscription tier of trusted children's programming expands, and parents pay slightly more for a show built by people who study how three-year-olds actually learn. The alternative version delivers a fully machine-generated baby feed, an endless scroll of inexpensive wonder tuned to hold a toddler's attention and sold for a fraction of a cent⁸. Children raised on that second version learn to expect the world to resolve itself, the way the boy expected his robot to.

Here is the part the numbers omit. Fun is how a child rehearses being a person. The waiting, the guessing, and the small victory each build something a feed will struggle to install afterward. Trade all of it for a frictionless scroll, and the child acclimates to ease and stays hungry for more of it.

What the boy wants back is the doing itself, the waiting and the attempting and the victory his own hands earn. A tool that eliminates the effort hands him a smaller life and calls it a gift. Our responsibility is protecting the difficulty worth keeping. Hand a child a screen that asks something in return. Give the fun back.

Sources

1. Bloomberg, "YouTube Creators Find a New Consumer for AI Slop: Babies," Bloomberg, Dec. 3, 2025, https://www.bloomberg.com/news/articles/2025-12-03/ai-slop-youtube-videos-for-kids-pretend-to-be-educational
2. Arijeta Lajka, "Investigation: YouTube Recommends AI Slop Videos to Toddlers and Preschoolers," The New York Times, 2026, https://www.nytimes.com
3. Fairplay, "Open Letter to YouTube and Google on AI Slop Aimed at Children," Fairplay, March 2026, https://fairplayforkids.org/wp-content/uploads/2026/03/YouTube-Letter-AI-Slop.pdf
4. Fortune, "200-Plus Organizations Urge YouTube and Google to Protect Kids," Fortune, April 1, 2026, https://www.fortune.com/2026/04/01/ai-slop-200-organizations-letter-youtube-google
5. MIT Technology Review, "How I Learned to Stop Worrying and Love AI Slop," MIT Technology Review, Dec. 23, 2025, https://www.technologyreview.com/2025/12/23/1130396/how-i-learned-to-stop-worrying-and-love-ai-slop/
6. Common Sense Media, "The 2025 Common Sense Census: Media Use by Kids Zero to Eight," Common Sense Media, 2025, https://www.commonsensemedia.org/research/the-2025-common-sense-census-media-use-by-kids-zero-to-eight
7. JAMA Pediatrics, "Screen Time at Age 1 Year and Communication and Problem-Solving Developmental Delay at 2 and 4 Years," JAMA Pediatrics, 2023, https://jamanetwork.com/journals/jamapediatrics/fullarticle/2808593
8. Deloitte, "Technology, Media & Telecommunications Predictions 2026," Deloitte, 2026, https://www.deloitte.com/us/en/insights/industry/technology/technology-media-and-telecom-predictions.html
9. Europol, "Facing Reality? Law Enforcement and the Challenge of Deepfakes," Europol Innovation Lab, 2022, https://www.europol.europa.eu/publications-events/publications/facing-reality-law-enforcement-and-challenge-of-deepfakes