Here’s a number worth sitting with: in a new study, every time researchers told parents that 10% more of their kid’s peers were using AI for homework, parents’ willingness to pay for an AI subscription went up by $1.83. Move the assumed peer-usage rate from 20% to 80%, and willingness to pay jumps more than 60% — for the exact same product, with the exact same features, doing the exact same thing.
Nothing about the AI changed. Only what parents believed other families were doing.
The study is a working paper out of the University of Chicago’s Becker Friedman Institute, by economist Leonardo Bursztyn — "Social Dynamics of AI Adoption in Parents’ Educational Decisions," forthcoming in PNAS. Researchers ran incentivized experiments with roughly 2,000 parents across the US, Canada, and the UK, showing them different (randomly assigned) scenarios for how many other teenagers were already using AI for schoolwork, then measuring what parents would actually pay for an AI subscription for their own kid.
The peer-pressure effect was strong and consistent. But the part worth pausing on is what happened next.
The researchers also tested whether giving parents honest information about AI’s risks would change anything — the kind of "here’s what to watch out for" content that shows up in most parenting advice. It didn’t move individual demand at all. Parents who learned about the downsides didn’t buy less AI for their own kid. What the risk information did change was their opinion about AI in schools generally — it made them more likely to say AI should be restricted or banned district-wide.
Read that twice. Parents wanted the guardrails for other people’s kids. For their own, the calculus didn’t budge.
The researchers call this a rat-race dynamic: everyone would prefer collective restraint, but nobody wants to be the one family that opted out while everyone else opted in. It’s the same logic that drives youth sports specialization at age six, or the private-tutor arms race in competitive school districts. The individual decision and the collective preference point in opposite directions, and information alone doesn’t reconcile them.
This matters for how Auxesis talks to parents, because it means the standard advice — "here’s a checklist to evaluate AI tools before you buy" — is aimed at the wrong problem. A checklist assumes parents are short on information. This study suggests they’re not. They’re responding to a social signal, and no amount of "5 things to check first" changes a decision that was never really about the tool’s features to begin with.
What actually interrupts a rat race isn’t more information. It’s a different structure to evaluate against — one that doesn’t ask "is everyone else doing this?" but asks something closer to "is my kid actually understanding more, or just finishing faster?"
That’s the whole premise behind the Parent Track. It’s not a list of AI red flags to memorize. It’s a sequence — Concrete, Pictorial, Abstract; Socratic questioning instead of answer-giving; a simple mastery log instead of a grade — that gives you something sturdier to check your kid’s progress against than "what’s everyone else using." When the question shifts from "am I falling behind" to "can my kid explain this back to me," peer pressure has a lot less to grab onto.
None of this means AI tools are automatically bad, or that every parent buying one is making a mistake. It means the reason most families end up buying one has very little to do with whether it actually helps — and knowing that is the first real check, before any feature checklist.
What’s the actual reason you’d say yes to an AI tool for your kid? Not "everyone’s using it." The real reason. Tell us below.
Sources: NBER Working Paper w34488, "Social Dynamics of AI Adoption," Leonardo Bursztyn · phys.org coverage
Leave a Reply