// methodology
How the assessment decides.
No black box. Here's exactly how the AI Impact Assessment scores your work and sorts each task into one of four categories — and why you stay in control of every step.
// the five dimensions
Every task, scored five ways.
Each task you list is rated 1–5 on five dimensions. The assessment suggests a starting score for each — and you can adjust any of them. The scores combine into an overall read on how suitable the task is for AI.
Frequency
1 = rarely · 5 = many times a dayHow often do you do this task?
The more often a task recurs, the more value there is in automating or accelerating it. High frequency pushes a task toward AI.
Predictability
1 = always different · 5 = same procedure every timeHow standardized and rule-based is it?
Predictable, repeatable work is the easiest for AI to handle reliably. This is the strongest signal toward AI Only.
Complexity
1 = mechanical · 5 = deep problem solvingHow cognitively demanding is it?
Higher complexity nudges a task back toward you — not because AI can't help, but because more of your thinking stays in the loop.
Risk
1 = trivial · 5 = serious consequencesWhat's the cost of getting it wrong?
High-stakes tasks call for human oversight. Risk pulls a task away from full automation and toward collaboration or human ownership.
Human judgment
1 = none · 5 = irreplaceableHow much trust, discretion, or relationship does it need?
Work built on trust, taste, and relationships is distinctly human. This is the strongest signal toward Human Only.
// from scores to categories
How the buckets are decided.
Frequency and predictability push a task toward AI — work that's routine and repeatable is the easiest to hand off. Risk and human judgment push a task away from AI and toward you, with complexity adding a gentler pull in the same direction.
The assessment weighs those signals together and places each task on a spectrum from "fully automatable" to "distinctly human," then sorts it into one of the four categories below. Because you can edit every score, you can see the category shift in real time as you refine your answers.
AI Only
Frequent, predictable, low-risk tasks that need little human judgment. These can largely run themselves — the plan is to set up the right tools and automations so you stop doing them by hand.
AI Assisted
Tasks where AI makes you meaningfully faster while you stay in the driver's seat. The plan is to learn the right AI assistants and the prompting skills to get good results.
AI Collaboration
More complex or higher-stakes tasks where you and AI agents work side-by-side, each doing what they're best at. The plan is to explore emerging agent workflows for this kind of work.
Human Only
Tasks defined by judgment, risk, and human relationships. AI plays little or no role — the plan is to keep deepening the skills that make this work distinctly yours.
// a note on accuracy
This is a starting point, not a verdict.
The assessment is a structured way to think clearly about your work — not a definitive judgment about your job. You know your role better than any model does, which is why you control the task list and every score. Treat the results as a conversation starter for your own AI learning plan.
Your job description and results stay in your browser. Nothing is saved to a server or shared.
See it run on your own work.
Free, anonymous, about five minutes. Your data stays in your browser.
Start your assessment