The traffic-light tool: where AI doesn't belong
Describe one AI use you're considering and answer the questions. The classification uses only the criteria in the essay "Where AI Doesn't Belong": any red criterion makes the use red, and green requires every green condition to hold. Your answers aren't stored anywhere — they stay in your browser and in the link you choose to share.
The questions
- Is the cost of an error high?
- Is the action hard or impossible to reverse?
- Does the decision affect rights, health, money, or a career?
- Is it impossible to explain how the decision was reached?
- Is there nobody qualified to genuinely supervise the system?
- Is there meaningful business or human impact?
- Is the data sensitive?
- Is the output relatively easy to check?
- Is there a clear definition of a good result?
The classifications
- Green — AI can execute — Here you can let AI work and leave people with spot-checking and correction when needed.
- Yellow — AI assists, a person decides — AI can organize information, surface patterns, and recommend. It shouldn't make the call alone.
- Red — AI does not make the decision — Once a process is classified red, the question is no longer how to improve the model. It's what part of the process should reach it at all.
Eight questions for the board
- What decision or action does the system perform?
- What business value do we expect from it?
- Who could be harmed if it's wrong?
- Can the action be reversed?
- Can we explain how the result was reached?
- Who is able to check the quality of the output?
- Who is the business owner of the system?
- Under what conditions do we stop or shut it down?