College (18–22): The Professional Stakes Are Real

College students are living through the most significant labor market disruption in a generation, and they're doing it during the four years when they're supposed to be building the skills that will carry them. The stakes couldn't be higher, and the confusion couldn't be greater — professors don't agree on AI policy, employers send contradictory signals, and the tools themselves are changing faster than any curriculum can track.

What distinguishes students who will thrive in this environment from those who won't isn't AI access — everyone has access. It's AI judgment: the ability to direct AI toward meaningful outcomes, to evaluate what it produces, to contribute genuine insight that AI can't generate on its own. That's the skill gap that's opening right now, and it's widening every semester.

At this age, you're no longer parenting in the traditional sense. Your relationship is adult-to-emerging-adult, and the most effective conversations are peer-level: you're thinking through this together, not telling them what to do. That shift in register matters. If you approach AI with your college student the way you did at 14, you'll lose them.

Skill atrophy in professional formation years. Whatever field they're studying, the foundational skills of that field — legal reasoning, engineering problem-solving, design thinking, clinical judgment — are built through struggle and iteration. AI that removes the struggle doesn't build the skill. Students who use AI as a replacement rather than an accelerant may graduate with credentials but without competency.

Portfolio and credential inflation without substance. AI can produce impressive-looking work samples. Students who build portfolios primarily through AI-assisted output may impress in the hiring process and then struggle in the first 90 days of a real job. That cycle is already happening.

Data exposure in academic and professional contexts. College students are pasting sensitive information — research data, client details from internships, drafts of proprietary materials — into AI tools without thinking about data retention policies. This is a real legal and professional risk.

Outsourced identity. Some students are using AI not just for tasks but for decisions: what should I major in, what job should I take, how should I respond to this conflict. When you outsource the hard decisions that form your sense of self and your values, you arrive at 25 not knowing who you are or what you actually think. This is subtler than the other risks but possibly the most important.

Shift to peer-level conversation. Lead with curiosity about how they're thinking about it, not guidance about what they should do. "How are your professors handling AI in your department? What do *you* think is the right approach?" is a much more productive opener than any form of advice-giving.

Ask about the portfolio question directly. "When you're applying for jobs, how are you going to show employers what you can actually do versus what AI helped you produce?" This isn't accusatory — it's a real question that most college students haven't fully worked through. Thinking it through with you is valuable.

Talk about data hygiene once, seriously. Not as a lecture but as a practical heads-up: "I read that a lot of people paste sensitive stuff from work into ChatGPT without thinking about it. Have you thought about that?" One mention, peer-level, and you've planted the flag. They'll think about it.

Encourage selective difficulty. "Pick two or three skills in your field that you want to develop without AI assistance, because those will be the ones that make you genuinely excellent." This reframes AI boundaries as strategic choice rather than moral restriction, which lands much better at this age.

Start with: "What's one thing you want to be genuinely excellent at — not just good enough at — by the time you graduate?"

What to listen for: Whatever they name is a gift. That's the thing to build the conversation around. From there: "How are you making sure you're actually developing that, and not just producing outputs that look like it?" That's the question that matters. And if they don't have a clear answer, that's useful too — it means they haven't fully thought it through, and this is a good moment to do that together.

• Skill Builder: Meta-Learning — Learning from experience to improve judgment over time
• Skill Builder: Interdisciplinary Thinking — Drawing on frameworks across domains
• Skill Builder: Knowledge Strategy — Knowing how you know what you know
• Skill Builder: Steering & Calibration — Directing AI toward meaningful outcomes
• Article: What We're Building Toward