Career Pathways

AI is restructuring the labor market. Some jobs are disappearing; many are being redefined; a few are being invented. Here’s how to think about where you want to land.

Jobs that didn’t exist five years ago

• Prompt engineer. Design and test prompts for production AI systems. The pure version of this role is already fading as models get better — but the underlying skill (precise specification under ambiguity) is everywhere.
• AI red-teamer. Paid to try to break frontier models. Hot demand, heavy psychological load.
• Data curator / annotator. The quality of training data drives model quality. Domain-expert annotation is the bottleneck for many specialized models.
• AI trust & safety. Content policy, abuse detection, incident response. Adjacent to but distinct from alignment research.
• AI-assisted creative.Directors who prompt Runway and Sora. Authors who develop books with Claude. Designers who ship with Figma + v0. The work isn’t “let the AI do it” — it’s curating and taste under acceleration.

Jobs that AI makes dramatically leverage-rich

• Software engineers who use Claude Code / Cursor / Codex — shipping 3–5× as much.
• Lawyers who use document-analysis AI to review contracts in minutes.
• Teachers who personalize at scale.
• Analysts who extract signal from noisy data ten times faster.
• Founders running entire small businesses solo.

Skills that compound

1. Write well. All AI collaboration runs through language. Sloppy thinkers get sloppy output.
2. Evaluate output.Can you tell if the AI is wrong in your field? That’s the only durable moat.
3. Design systems, not prompts. The interesting work is wiring up AI to real workflows, not one-shot chat.
4. Domain depth. AI makes generalists cheap. Specialists who deeply understand medicine, law, engineering, education become more valuable, not less.

Where to go next in high school

• Build something small and public that uses an AI API. Ship it.
• Contribute to an open-source AI project.
• Enter an AI hackathon. Most colleges now run them.
• Take a statistics course. It’s the one class that underlies all of this.