What should children learn if AI can perform most knowledge work?
What should children learn if AI can do most knowledge work?
The reflex answer is “creativity and critical thinking,” which is comfortable and nearly useless as guidance. Here's an attempt at something more specific.
Ask what children should learn in an age of capable AI and you reliably get: creativity, critical thinking, emotional intelligence. It's not wrong. It's also the answer people gave for every previous technological shift, which is a hint that it's functioning as reassurance rather than analysis.
Let me try to be more specific, and accept being more wrong.
The trap in “don't learn what AI can do”
The obvious inference — stop teaching what machines do well — fails on inspection. Calculators didn't make arithmetic pointless to learn, because arithmetic fluency is scaffolding for algebra, and algebra is scaffolding for everything after.
The same logic applies here. Writing isn't valuable only as output; it's the mechanism by which people find out what they think. Outsource it early enough and you don't get a person with a writing assistant. You get a person who never built the underlying ability.
Four that look load-bearing
Reading something difficult and slowly. The capacity to hold a hard argument in your head long enough to find the weak joint. Summaries are a substitute for this exactly until the summary is wrong.
Writing to think. Not to produce documents — to discover what you believe. A child who can only edit generated text has skipped the step where the thinking happens.
Evaluating a claim. In an environment of cheap, fluent, confident output, distinguishing a well-supported claim from a well-phrased one becomes a primary survival skill rather than an academic one.
Sustained attention on something that isn't working. Possibly the scarcest of the four. Most valuable work is a long stretch of not-yet-working, and tools that always produce an immediate plausible answer remove the practice of enduring that.