AI career risk: how to tell if your job is actually exposed.

Most "AI will take your job" content is written for clicks. This page is written to help you make one honest call: is the way you earn a living materially more exposed to generative AI in the next 12 months?

The four signals that predict task-level exposure

1. Repeatable text/output. If your work produces reports, emails, summaries, code, or structured docs, that's the first place a model already competes with you.
2. Weak feedback loop. If nobody notices when your output drops 20% in quality, an AI draft is "good enough" to your employer.
3. No physical or relational moat. Tasks that don't require presence, negotiation, or trust are the most substitutable.
4. Tool-agnostic buyer. If your client doesn't care how the work gets done, the tool that does it faster wins.

The exposure grid

Score each of your top 3 tasks on those four signals. Anything scoring 3+ is High exposure. A single High-exposure task doesn't mean you'll lose your job — it means your employer will pay you less for that task unless you reposition around it.

The one move that reduces exposure fastest

Stop trying to be faster than the model. Instead: own the judgment layer. Rewrite one task per week from "I do X" to "I decide which X to ship." Judgment, taste, and accountability are the pieces your employer can't outsource to an API.

Kyrovo scores are practical signals, not certainty. Judgments about your career are yours to make.