Part 1
Start with tasks, not headlines
A job title can hide very different tasks. A systems analyst, policy analyst, bylaw officer, and nurse can all use software, but the replacement risk is not the same.
- List the tasks that are repeatable, digital, and document-heavy.
- List the tasks that require a person on site, licensed judgment, safety responsibility, or public trust.
- Treat AI as a workflow-change question first, not a simple job-death prediction.
Part 2
Higher-exposure signals
These signals do not mean a job disappears. They mean the work may change faster and entry tasks may be compressed.
- Routine forms, scheduling, email drafts, summaries, reporting, and data cleanup.
- Research scans, first drafts, basic coding help, ticket triage, and document comparison.
- Back-office workflows where software can reduce repetitive coordination work.
Part 3
Lower-exposure signals
Work that requires presence, accountability, physical response, public judgment, or regulated care is less likely to be fully replaced by a text or workflow tool.
- Driving, field inspection, utilities, emergency response, patient care, and hands-on technical work.
- Licensed decisions where a person or employer remains accountable.
- Conflict, care, discretion, public complaints, and safety decisions.
Part 4
How Make100K will score this later
Make100K does not publish numeric AI exposure scores yet. The data model is in place, but scores need sourced task evidence and manual review before they should be public.
- Every score should explain which tasks are exposed and which tasks are harder to automate.
- Scores should be out of 100 only after the rubric is reviewed against occupational task sources.
- A useful score should help people choose training and skills, not scare them.
Common questions
Can Make100K tell me exactly whether AI will take my job?
No. The useful question is which tasks are exposed, which tasks still need people, and what skills or paths reduce risk.
Why not show a score right now?
A numeric score would be misleading until it is backed by sourced task data and reviewed notes. The site currently shows the rubric structure and not-scored-yet status.