The question every parent is asking wrong
The anxiety is rational. You're raising a child who will enter the workforce around 2040, into an economy being rearranged by machines that improve monthly — and every instinct says prepare them, which the last generation's playbook translated as: pick the winning skill early. Coding bootcamps for nine-year-olds. Prompt-engineering summer camps. The arms race of enrichment.
Here's what the playbook misses: specific-skill predictions have a miserable track record — the safest career advice of 2010 ("learn to code") is precisely the work AI absorbed first — and the 2040 job market is less predictable than any before it. You cannot aim a child at a target that won't hold still. What you can do is ask a better question: when machines produce the cognitive output, what stays scarce — and how does a human actually acquire those scarce things between ages two and eighteen?
The first half is economics. The second half is developmental science. They agree with each other almost embarrassingly well.
What stays scarce (the economics)
The Great Inversion in one line: as cognitive output gets cheap, the premium migrates to what machines can't manufacture — judgment, trust, presence, taste, and the regulated human state underneath all four. Map that onto childhood and the "future-proof curriculum" writes itself:
- Self-regulation — managing one's own state, impulses, and recovery. The substrate of everything else.
- Sustained attention — increasingly rare, therefore increasingly valuable, in an economy spending billions to fragment it.
- Curiosity that survives — the drive to ask real questions and tolerate being a beginner, which is the entire engine of adapting to jobs that don't exist yet.
- Human connection — reading people, repairing conflict, earning trust. The relationship layer machines can simulate but not occupy.
Notice what this list is: the developmental psychology syllabus that existed before AI was a word. The famous longitudinal work — Dunedin's thousand children tracked for decades, the delay-of-gratification literature in its modern, corrected form — keeps finding the same thing: childhood self-regulation predicts adult health, finances, and stability better than IQ or social class across a startling range of outcomes. The machines didn't change what makes humans capable. They just removed everything else from the market.
The safest preparation for an unpredictable economy was never a skill. It was a nervous system that can meet whatever arrives — and that, unlike coding, has a critical period.
How children actually acquire it (the science)
Now the second half, where parenting advice usually goes wrong. None of the four capacities above is installed by instruction. You cannot lecture a seven-year-old into self-regulation. The acquisition channels, per the research, are:
- Co-regulation before self-regulation. A child's stress system calibrates against the adult systems around it — literally. Thousands of moments of being soothed, met, and settled by a regulated adult become, over years, the internal machinery of settling oneself. The developmental sequence is non-negotiable: borrowed calm precedes owned calm.
- Observation over explanation. The social-learning and mirror-neuron literature is blunt: children copy what's modeled with far higher fidelity than what's taught. They're not listening to your words about patience; they're recording your behavior in traffic.
- Productive struggle. Tolerance for difficulty grows only through metabolized doses of difficulty — the puzzle not solved for them, the boredom not rescued by a screen, the failure followed by another attempt. Every instant rescue is a repetition of the lesson discomfort is unbearable.
- Unstructured time. Boredom isn't a gap in the schedule; it's where self-directed thought, imagination, and the ability to generate one's own agenda develop. A childhood programmed to the minute outsources the very executive function it's trying to build.
The uncomfortable part: you are the curriculum
Put the two halves together and the conclusion lands somewhere most parenting content is too polite to go: the decisive variable isn't the child's enrichment schedule — it's the adult's state.
The parent checking their phone 150 times a day is running an attention masterclass, enrollment automatic. The parent who melts at every setback is teaching the stress response by demonstration. And the parent who lectures about screens while scrolling at dinner is teaching something even deeper: that stated values and lived behavior are separate things — which children absorb perfectly and forgive slowly.
This is hard news with a generous flip side: it means the highest-leverage investment in your child's 2040 is your own regulation, attention, and relationship with technology — all trainable, all improvable, and all of it counts double because it compounds in two nervous systems at once. (The adult curriculum is this entire site: regulation, attention, the phone.) Children become what you are. Not what you say. Not what you schedule. What you are, on an ordinary Tuesday, when you think they're not watching. They're always watching — it's their job.
Stop asking "what should my child learn?" and start asking "what am I demonstrating daily?" The first question buys camps. The second changes households — and the household, not the camp, is where 2040's adult is being assembled right now.
The practice: six moves
- Regulate yourself first — visibly. Let them see you take the slow exhale before responding, name your frustration calmly, recover from a bad day out loud. You're not hiding the struggle; you're modeling the repair. That demonstration is worth a hundred lectures.
- Protect boredom like an asset. Some hours of every week should have nothing scheduled and nothing screened. The whining phase passes; what follows it — self-generated play, actual thinking — is the product.
- Let the struggle finish. When the homework frustrates or the tower falls, the move is presence without rescue: "that's hard. I'm here. keep going." Tolerance for difficulty is built in exactly these doses, and nowhere else.
- Make meals a device-free protocol — for everyone. The research on family meals keeps finding outsized benefits, and the mechanism is simple: it's the daily session where conversation, attention, and connection get practiced. The "for everyone" clause is the entire policy.
- Treat AI as a thing you do together, with judgment running. Don't ban it (it's their world) and don't hand it over raw. Sit together: ask it things, catch its mistakes, talk about when it's brilliant and when it's confidently wrong. You're teaching the relationship — tool, not oracle — which is the one tech lesson that transfers to every future tool.
- Audit the patterns you're handing down. Most of what travels between generations travels through behavior — the stress responses, the conflict styles, the relationship with rest you absorbed and now demonstrate. The chain breaks at awareness, and breaking it is the most consequential parenting work there is.
So… coding, screens, and AI tools?
The fast verdicts, with their honest sizes: coding — excellent logic training, fine as passion, no longer a moat; treat it like music lessons. screens — dose and displacement matter more than the panic suggests, and your modeling matters more than your rules. AI fluency — genuinely useful, genuinely easy; a regulated, curious, socially-capable child acquires any tool in weeks. The reverse — a tool-fluent child without the substrate — is precisely the profile the inversion punishes.
The whole answer, compressed: build the human; the human handles the tools. It's also, not coincidentally, the answer to most questions this century is about to ask.
The full parenting system, built on the science.
NextGen is the complete protocol — co-regulation, attention, patterns, the AI-age curriculum — for raising humans who'll thrive in 2040. Includes 3 months of Marsa Coach.
See NextGen →Frequently asked questions
What skills should kids learn for the AI age?
The machine-proof capacities: self-regulation (the strongest childhood predictor of adult outcomes), sustained attention, curiosity that tolerates being a beginner, and human connection. Tools are the easy layer on top — the substrate is the preparation.
Should kids still learn to code?
As logic training and literacy, yes; as a career moat, no — AI produces code now, and the premium moved to judgment about what to build. Treat it like music lessons: great for the mind, not a guarantee.
How does parental behavior affect children's development?
It is the curriculum: children learn by observation and state-matching, and their stress systems calibrate against yours (co-regulation). Your phone habits, failure responses, and ordinary-Tuesday state outweigh anything you schedule.
How do I prepare my child for jobs that don't exist yet?
Optimize capacities, not job guesses: protect boredom, allow productive struggle, model your own regulation and tech relationship, keep device-free family attention — and use AI together, with judgment switched on.