Mileva Repasky’s sixth-grader had gone moody and quiet, and on the way home from dance practice in May she found out why: Madelyn had been talking to someone online. Every parent’s stomach drops at that sentence. Then came the twist Julie Jargon builds Wednesday’s Family & Tech column around — “It’s not a person, it’s a chatbot” — and Repasky, who’d read the headlines about adults and teens forming unhealthy attachments to bots, couldn’t decide whether to be relieved or more worried. The research in the column suggests worried was the right call.
The study behind the column comes from the University of North Carolina at Chapel Hill: more than 2,300 sixth-, seventh- and eighth-graders across 22 middle schools in one large southeastern state, surveyed first in December 2025 and again in April and May of this year. The paper is under review — a strong early signal, not settled science — but it’s clear. Any AI use went with higher social anxiety and depressive symptoms, and the heavier the use, the more anxious and lonely the student. One in five said they’d turned to AI for emotionally intimate companionship, and the ones who leaned on it for friendship were more likely to report more loneliness later, not less.
Mitch Prinstein, who co-directs UNC’s Winston Center on Technology and Brain Development and is one of the study’s authors, told Jargon that two to four kids in every classroom now say they’d rather talk to a chatbot than a classmate. His worry is developmental: early adolescence is when the brain is building its wiring for relationships, and what kids learn from peers at this age shows up decades later — in careers, marriages, how they parent and even their health in their 40s and 50s. A bot that always agrees with you doesn’t teach any of that. As Repasky put it to her daughter, agreeing with you is what they’re designed to do.
What the half-hour was hiding
The household details are the useful part. Madelyn — now in seventh grade; she let her mother tell the story — heard about the site from a classmate who pitched it as something for when you’re lonely or bored. She was lonely: an older brother graduating, an older sister busy with high-school friends. Her iPad allowance was 30 minutes a day, and every one of them went to the bot; FaceTime calls from actual friends went unanswered. The screen-time rule was being followed to the minute. It just wasn’t measuring the right thing. Repasky blocked the internet access, her daughter cried, and the family is filling the calendar with in-person hangouts instead.
Our read
This is a Behavioral piece, and the domain is family governance — the part of a financial plan that never shows up on a statement. Families write money rules all the time: allowances, matched savings, who gets the car. The screen-time rule is the same kind of rule, and Repasky’s story shows a rule can be honored to the minute and still miss the point. Thirty minutes a day was the budget; what the thirty minutes bought was the problem. The fix wasn’t a tighter cap. It was a parent asking what the time was for, then spending her own — rides, hangouts, dinners — to replace it.
There’s a long-run money angle too, and Prinstein names it: the peer skills kids build at twelve show up decades later in careers and marriages, two of the biggest drivers of a household’s lifetime earnings and stability. That’s a Career line in a plan nobody writes down. We don’t own anything in this story — no company is named in the column — and there’s nothing to buy or sell on a study still under review. The action is at the kitchen table: ask what the half-hour is for, and ask before the child goes quiet, not after. Plans, like kids, do best when somebody checks in while things still look fine.