Raising an Adaptable Generation: What to Actually Tell a Teenager About AI and Careers

"Pick something safe" was genuinely reasonable career advice to give a teenager in 1995, when safety mostly meant choosing a stable, established profession over a risky, unproven one. Federal Reserve Bank of New York data on recent college graduates suggests that framing may now be exactly backwards, and that some of the fields that feel safest to a worried parent are quietly among the most exposed.

What the Data Actually Shows

New York Fed research tracking underemployment among recent graduates has found that certain established, credential-heavy majors, ones parents have steered kids toward for decades specifically because they seemed safe, now carry meaningfully elevated underemployment rates, as entry-level tasks in those fields prove to be exactly the kind of structured, well-defined work current AI systems handle competently. Meanwhile, World Economic Forum data shows a roughly twenty-three percent wage premium already attached to explicit AI-skill competency across a wide range of fields, safe and unsafe alike. The entry-level rung, historically the easiest step onto any career ladder, is precisely what’s compressing fastest, because that rung has always been built from exactly the kind of bounded, repeatable tasks AI systems now perform well.

Why "Safe Major" Was Always a Proxy, Not the Real Target

The advice to pick a safe major was never really about the major itself. It was a proxy for something parents actually wanted: durable employability, a reasonable income, work that wouldn’t disappear. The proxy made sense when major and employability were tightly correlated. That correlation has weakened considerably, which means the advice needs to target the actual goal directly, adaptable capability, rather than continuing to optimize for a proxy that’s drifted away from what it used to predict.

What to Actually Recommend Instead

Two shifts matter more than which major gets chosen. First, weight the twelve human superpowers, trust, leadership, empathy, judgment, negotiation, relationship-building, taste, purpose, accountability, creativity, complex thinking, adaptability, as seriously as technical coursework, because these are precisely the capabilities that remain valuable regardless of how any single field gets reshaped by AI. Second, build minimal, practical AI fluency before the first job offer rather than after, learning to work alongside these tools as a baseline skill, the way basic computer literacy became a baseline skill a generation earlier, not as an optional specialization for the technically inclined.

A Different Conversation to Have

Rather than asking a teenager "what field feels safe," a more useful question is "what combination of durable human capability and practical AI fluency would make you valuable in more than one field." That reframes the goal away from picking the one correct destination, which nobody can reliably predict a decade out, toward building a transferable foundation that holds up regardless of which specific fields grow or shrink.

Starting This Conversation

If there’s a teenager or young adult in your life making these decisions now, the most useful thing you can offer isn’t a confident prediction about which field will still exist in fifteen years, nobody actually has that. It’s help identifying which of the twelve human superpowers they’re already naturally strong in, and one concrete way to start building genuine, hands-on AI fluency this year, not as an elective, as infrastructure.

Recommended Reading

This article draws directly on the book this Sunday series is built around:

• AI Will Steal Your Job – Here’s How to Steal It Back, by Palatino D. James

Browse the full Strong Through Change library →