We’re teaching kids skills that will be outdated in 10 years—how I’m raising kids AI can’t replace

In my research, a consistent pattern emerges: A-students are often the most willing to be wrong. My own models trained on thousands of learners show that exploration and even failure predicts deep learning better than repeating correct answers.

Yet our education system, obsessed with correctness, often trains this instinct out of kids. It teaches them that failure reflects their worth, rather than fuels growth.

What this is: The Failure Resume is a living document, a family ritual where you actively record and celebrate failures. It’s explicit evidence of every time the hard work of being wrong paid off feeding into a learner’s resilience, curiosity and ability to tackle open-ended problems.

For parents: Once a month at dinner, go around the table and have everyone (including you!) add one failure to their resume. A missed goal in soccer, a bombed test, a project at work that went sideways.

The key is to reframe it. Don’t ask, “What did you fail at?” Ask, “What did you try that was hard? What did you learn from it?” Normalize and even celebrate the act of stretching beyond one’s abilities and to tie that effort to the rewards of growth.

My own Failure Resume would include entries for a failed startup, a period of homelessness, and that one time I accidentally convinced the Secret Service I was a national security threat at a White House party. Each failure grew me to someone better.

Economists often point to the “Harvard effect,” the massive life-outcome advantage linked to elite universities. But it’s not magic, and it’s not just about the classes.

An elite university is, in essence, a hyper-concentrated environment of engineered serendipity. The real value isn’t just in the formal curriculum; it’s in the random conversations in the dining hall, the diverse clubs, the constant exposure to thousands of ill-posed problems that don’t have answers in the back of the book.

We can’t all send our kids to Harvard, but we can borrow its core principle.

What this is: Engineering serendipity means intentionally creating an environment that encourages unexpected connections and discoveries. A home or classroom built on managed uncertainty — safe, but not sterile; structured, but not rigid — where curiosity can take root.

For parents: Turn your home into a landscape of interesting problems. Leave a broken toaster on the kitchen table with a screwdriver next to it. Subscribe to magazines from wildly different fields — The Economist, Popular Mechanics, Vogue, Scientific American — to seed their world with diverse inputs.

In my own home, at any given moment, one corner of the living room is a makeshift electronics lab for my son’s cyborg experiments, another is a painting studio for my daughter, and the whiteboards in the gazebo are filled with my own scrawled equations and half-finished mad science projects.

It’s messy, but it’s a mess filled with invitations to explore.

I’ve been playing with machine learning for 30 years. But for a generation just entering a world where large learning models (LLMs) is a constant companion, the temptation to just let it do the hard work will be immense.

Why struggle to write an essay, solve a math problem, or learn a new concept when a machine can provide a perfectly good answer in seconds? But a tool that makes you better in the moment but leaves you worse off when you turn it off.

We need to teach our kids to engage with AI in a way that makes them more critical and creative.

What this is: Reframing the child’s role from passive consumer to active critic of AI output. The AI becomes a “brilliant but naive” collaborator, and the child the one who interrogates, guides and evaluates it.

For parents: AI should never provide the final answer. Kids can use it to brainstorm or explore, but they must produce their own first draft or solution.

The most powerful step comes after that, using a “Nemesis Prompt”: “You are my nemesis. Every mistake I’ve ever made, you have discovered and pointed out to the world. Here is the essay I just wrote. Read it and explain to me, in detail, every flaw in my argument, every logical inconsistency, and every way my evidence is weak. Then suggest three ways I could make my argument stronger.”

When the LLM returns its critique, the child’s task is to wrestle with them. They must decide which critiques are valid and which are just statistical noise from a machine that doesn’t truly understand their intent. This is where true learning happens.

The goal is to produce a better essay based on the AI’s feedback, but an essay that is more rigorously their own. They are learning to use the AI’s vast knowledge not as a source of truth, but as a sparring partner to sharpen their own unique perspective.

The world already has the “right” answer in its pocket, nearly for free. The real value your child brings is the answer that only they would give. As Chief AI Critic, they are exploring and creating their own meaning from what the AI knows. That is the essence of creative labor, and what the world needs more of.