How to build an AI-proof career by chasing genuine curiosity

Executive overview

Over 50% of workers are disengaged, and AI is accelerating the cost of being in the wrong career. The jobs most at risk are not the ones people feared — white-collar roles built on language and routine pattern-matching are already being disrupted.

The defence is not to resist AI but to become its most capable operator. Curiosity is the compounding advantage: people who can't stop learning in their field outpace both peers and models, because edge knowledge hasn't been written down yet.

The highest-risk move in an AI world is playing it safe in a job you don't love.

Why "safe" careers are now the most exposed

• Jobs built on language — translation, paralegal work, text processing — are the earliest casualties
• Disengaged workers are sitting ducks: they're not pushing to improve and they're not leaning into new tools
• AI anxiety can freeze people into inaction, which is worse than any individual career risk
• The dystopian AI narrative is five times more prevalent in the US than in China — treat it as a spur, not a signal to stop

The six principles (Gurley's framework)

1. Chase your curiosity — the career you'd study for free on weekends is the one worth pursuing
2. Hone your craft — continuous learning compounds; treat finishing school as the start, not the end
3. Give yourself permission — most people never pursue what they want because no one told them they could
4. Strong opinions, loosely held — unlearn what made you successful when the evidence changes
5. Build aspirational mentors — create digital profiles of people you admire, study them like a fanboy; tests whether the field genuinely excites you
6. Build a peer group — four to six like-minded people on the same journey, outside your org, sharing ideas regularly

How to identify genuine curiosity

• Look at hobbies and weekends — what do you do without being asked?
• Ask annually: "Do I see myself doing this in 30 years? How do I feel about that?"
• Use Bezos's regret minimisation framework: what would your 80-year-old self advise?
• If studying the field sounds tedious, it's the wrong field; if you can't stop, you're in the right lane
• Edge knowledge — things being discovered now that aren't yet in any model — is where humans still have a decisive lead

Becoming the most AI-enabled version of yourself

• The types of prompts you think to give AI increase the more you use it — lean in to learn what it's capable of in your field
• Use it to research, prototype, and ideate simultaneously
• Build Claude or ChatGPT projects around people you want to learn from: dump books, interviews, and transcripts
• Before asking a mentor a question, ask AI first — it can simulate the knowledge base of almost anyone well-documented
• Coders: be the person who knows how the new AI tools work, not the one grinding out routine code

Finding and using mentors and peers

• Most people shoot too high on mentors and get no response; come down a rung or two
• Start with a small ask: "I'm facing this choice — any thoughts?" not "Will you be my mentor?"
• Build aspirational mentor profiles first (podcasts, books, YouTube, AI summaries) — it builds confidence and tests your genuine interest
• For peers: create a WhatsApp group, Slack, or Google Space with four to six people on a similar journey outside your organisation
• Peer groups surface job opportunities, provide perspective on bad days, and help distinguish "wrong company" from "wrong career"

What parents and career counsellors get wrong

• Optimising for economic safety pushes people into roles they don't love — and those are now the most exposed to AI
• Over-scheduling children for college applications (chess, lacrosse, volunteering) leaves no room for genuine exploration
• Specialising into a major before finishing sophomore year removes the chance to discover what you actually enjoy
• Better approach: maximise exposure to different things, watch for signs of genuine interest, don't force early answers

One action to take this week

• Pick two or three potential new directions and build a "battle card" for each: role-play the transition, ask AI to sketch out the first week
• Fill in details over time — the abstract option you keep returning to is probably the right one
• Do not stay frozen; scenario planning with AI costs nothing and breaks the paralysis