Skip to main content

    Free tool

    Should we use AI for this task?

    How do you decide whether a task is a good fit for AI?

    Look at two things separately. Fit: is the task repetitive, frequent, hard to solve with simple rules, and measurable? Risk: is the data available and allowed, what happens when the AI is wrong, will a person review the output, and must decisions be explained? A task with high fit and low risk is a strong candidate; high risk means a reviewed pilot first; missing data means not yet.

    Answer nine questions about one task

    1. 1. How repetitive is the task?
    2. 2. Could clear rules solve it instead?
    3. 3. Is the data it needs available, and are you allowed to use it?
    4. 4. What happens if the AI gets it wrong?
    5. 5. Will a person review the outputs?
    6. 6. How often does the task happen?
    7. 7. Must decisions be explainable to customers or regulators?
    8. 8. Is there a success metric?
    9. 9. Which best describes the work?

    Answer all nine questions to see the verdict. (0 of 9 answered)

    Frequently asked questions

    How is the verdict worked out?

    Each answer adds points to AI fit (up to 15) or risk (up to 10). Go needs fit 9 or more and risk 3 or less; Pilot first needs fit 6 or more and risk 6 or less; anything else is Not yet. If the data is not available, the verdict is always Not yet.

    Why does it sometimes recommend not using AI?

    When clear rules can solve a task, rules-based automation is usually cheaper, faster and easier to audit than AI.

    Is my answer stored?

    No. The triage runs in your browser and nothing you enter is sent to us.

    More free tools

    Assess AI ReadinessContact