What ChatGPT Changed About What Kids Need to Learn.

In a couple of years, the answer to almost any school question became available in a second. A parent's natural reaction is anxiety: "so what do we even teach now?" And right behind it comes a tempting conclusion: "no more memorizing – we'll just ask the AI." That conclusion is worth unpacking, because it sounds logical but turns the science of thinking upside down.

In short: what actually changed

What changed isn't the list of things worth knowing – it's the balance. Recalling a fact quickly has lost value; a machine fetches it. What's gained value is everything the machine doesn't do for you: asking a precise question, judging and verifying the answer, spotting the error, connecting one thing to another, deciding what to do next. But – and this is the key – all of that runs on knowledge you already have. So the honest conclusion isn't "facts don't matter," it's "facts are no longer the finish line; they're the foundation."

What the skills data says

The World Economic Forum surveys employers each year on which skills matter most. In the 2025 report, the top core skill is analytical thinking, followed by resilience and flexibility, leadership, creative thinking, and motivation and self-awareness; technological literacy and curiosity and lifelong learning are also in the top tier. Employers expect 39% of core skills to change by 2030.

Separately, the WEF ranks which skills are growing fastest in demand – there, AI and big data lead, followed by cybersecurity and technological literacy, with creative thinking and curiosity right behind. The distinction matters: analytical thinking is the most needed skill now; AI literacy is the fastest-growing one.

The main shift: from "knowing the answer" to "judging the answer"

Language models sound confident even when they're wrong – inventing facts, links, and quotations. So a distinct competency has appeared: AI literacy – the ability to critically evaluate a machine's answer rather than take it on faith (see UNESCO's competency framework, 2024; Long and Magerko, 2020).

Put simply, the work has moved: what once mattered was finding information; now it's verifying it. What used to be valued was remembering; now it's noticing that an answer is elegant and wrong.

A caution: "cognitive offloading"

There's a flip side. In Gerlich's study (2025, Societies, 666 participants), frequent use of AI tools was associated with lower critical-thinking scores, and the link was explained by "cognitive offloading" – the habit of handing thinking over to the machine. The effect was more pronounced among younger people (17–25). Microsoft Research and Carnegie Mellon found something similar (Lee et al., 2025): the more people trusted the AI, the less they scrutinized it, and the work itself shifts from searching to verifying and "overseeing."

Honest caveats (important): both studies are correlational, on adults, and largely self-reported. This is not proof that "AI damages a child's brain," and the causality could run the other way (those less inclined to think may delegate more). But the signal is clear: the danger isn't the tool – it's the habit of not thinking.

But knowledge hasn't gone anywhere

Here's the counterintuitive part. Critical thinking isn't a free-floating skill you can apply to anything. It runs on what's already in your head. Cognitive psychologist Daniel Willingham puts it plainly: you cannot develop "thinking skills" in the absence of facts.

Which leads to a simple consequence: to ask the AI a good question and catch it in an error, you have to know something about the topic yourself. "Just ask ChatGPT" fails for exactly this reason: you get a confident answer with nothing to check it against. The machine took over the searching – it didn't take over the need to understand.

What to teach now

Lost value	Gained value
remembering a fact you can look up in a second	knowing enough to spot an error in the answer
the single "right answer"	the ability to ask a precise question
retelling and cramming	verifying: "how do we know? is that really so?"
speed of searching	judgment: what to do with this, is it ethical
following a template	transferring knowledge across fields and creating something new

What a parent can do

• Think first, then ask. Have your child form their own version before opening the AI. That protects their own thinking.
• Teach verification. Three questions for any answer: how is this known? could it be otherwise? what here might be untrue?
• Argue with the machine together. Ask the AI, then hunt for the weak spot in its answer. A great game and the best lesson in AI literacy.
• Build the base. Reading, conversation, observation – everything that puts knowledge in their own head. Without it there's nothing to check against.
• Value the question above the answer. Praise a good question as much as a correct answer.
• AI is a conversation partner, not an oracle. "Let's ask and check" instead of "let's find out the truth."

Frequently asked questions

So kids don't need to learn facts anymore? They do – but for a different reason. Not to shine in a quiz, but to have something to think with: to understand the question, notice an error, judge the answer. Facts became the foundation, not the finish line.

Does AI harm a child's thinking? There's no direct evidence: the studies are correlational and on adults. What harms isn't the tool but the habit of handing thinking over to it. The rule "your own version first, then the machine" helps.

At what age should a child use AI? First the base and the habit of thinking; AI later, as a tool with an adult nearby, with a conversation about the fact that it makes mistakes. Age matters less than whether the child has anything to check the answer against.

Will AI make school obsolete? No. Both the WEF and the OECD describe not the disappearance of education but a shift: more AI literacy, more higher-order thinking, and more "human" skills – judgment, creativity, and learning how to learn.

How Wings helps

Wings trains exactly what has gained value. Every day, one short task or question for the child (ages 2–14) with no single right answer: they have to reason, give an argument, look from another side. We grade not correctness but the quality of reasoning – the very thing that separates a person who checks the machine from one who copies its answer. Five minutes a day, screen-free and pressure-free. Try it free →

Sources

• World Economic Forum. Future of Jobs Report 2025 (January 2025). (Analytical thinking – top core skill; 39% of core skills to change by 2030; separate ranking of fastest-growing skills led by AI and big data.)
• World Economic Forum. Future of Jobs Report 2023. (44% of core skills set to change within 5 years; largest demand growth for creative thinking.)
• UNESCO (2024). AI Competency Framework for Students. (12 competencies; critical judgement of AI solutions.)
• Long, D., & Magerko, B. (2020). "What Is AI Literacy? Competencies and Design Considerations." Proc. CHI 2020.
• Gerlich, M. (2025). "AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking." Societies, 15(1), 6. (666 participants; correlational, not causal.)
• Lee, H.-P., et al. (2025). "The Impact of Generative AI on Critical Thinking…" Proc. CHI 2025 (Microsoft Research / Carnegie Mellon). (Self-reports from 319 knowledge workers.)
• Willingham, D. T. (2009). Why Don't Students Like School? (Thinking depends on knowledge held in memory.)
• Anderson, L. W., & Krathwohl, D. R. (2001). A Taxonomy for Learning, Teaching, and Assessing (revised Bloom's taxonomy).
• OECD (2019). Learning Compass 2030. (Creating new value, reconciling tensions, taking responsibility.)

Last updated: July 2026.

Try it free

Leave your email and pick your child's age. For the next 5 days you'll get one WINGS task a day – matched to their stage, straight to your inbox.