What's changed: Deepened AI-900 Chapter 5 to the AZ-900 baseline (pretraining + three tuning levels prompt/RAG/fine-tune, generative model types, Azure OpenAI models/Studio/embeddings, Copilot product examples, Content Safety/prompt injection, FAQ)
5.3Responsible Use and Risks of Generative AI
Understand generative-AI-specific risks (hallucination—plausible but wrong output, bias, copyright/privacy) and how to use it in line with Responsible AI principles.
Generative AI is powerful but comes with errors and risks. Beyond convenience, understand its limits and use it per the Responsible AI principles (Chapter 1). Here we cover risks specific to generative AI and how to mitigate them.
5.3.1Generative-AI-specific risks
- Hallucination: confidently generating plausible but incorrect content; fact-checking is needed.
- Bias: bias in training data appears as biased output (relates to fairness).
- Copyright/IP: generated content may resemble existing works; including others’ works in input is risky.
- Privacy: putting personal/confidential data in prompts risks leakage.
- Harmful content/misuse: inappropriate output or jailbreaks (bypassing limits); mitigate with content filters.
Continue reading — free sign-up
You're reading the free preview. Sign up free to read this section in full, plus every chapter (including 4+) and all questions.