Matching Exercise: Risk Levels in the use of AI