2. Principles of Responsible AI

Microsoft frames responsible AI as six principles, and the exam expects you to recognize which principle a scenario is about. Here's the thing that makes this phase easy if you understand Phase 1 and hard if you only memorize: each principle is a direct response to a real limitation of how AI systems work. Models learn patterns from data, so they can inherit bias (fairness). They predict rather than know, so they can fail unpredictably (reliability and safety) and need humans answerable for them (accountability). Keep asking "which limitation does this principle guard against?" and the six stop blurring together.