Azure AI Fundamentals

Sample AI-901 practice questions

5 of the 246 original questions in our AI-901 bank, written to match the real exam's style and difficulty. Every question includes a full explanation — expand each answer to check your reasoning.

1. Question 1medium

   An organization wants to ensure its AI systems treat all demographic groups equitably and do not produce discriminatory outcomes. Which Microsoft Responsible AI principle does this primarily address?

   • AReliability and safety
   • BFairness
   • CTransparency
   • DPrivacy and security
   Show answer & explanation

   Correct: B

   Fairness is the Responsible AI principle concerned with AI systems treating all people equitably and avoiding discriminatory impact across demographic groups. Reliability and safety is about consistent, safe performance; transparency is about understandability of AI decisions; and privacy and security is about protecting data. Microsoft’s six principles (fairness, reliability and safety, privacy and security, inclusiveness, transparency, accountability) are heavily tested on AI-901.

2. Question 2hard

   A developer wants to ground a generative AI chat application in their company’s own product documentation so answers stay accurate and current. Which approach on Microsoft Foundry is the best fit?

   • AFine-tune a base model on every customer conversation nightly
   • BImplement a retrieval-augmented generation (RAG) pattern that retrieves relevant documentation and passes it to the model as context
   • CIncrease the model’s temperature setting
   • DSwitch to a smaller model to reduce latency
   Show answer & explanation

   Correct: B

   Retrieval-augmented generation (RAG) is the standard pattern for grounding a model in private, current knowledge: relevant chunks of the company docs are retrieved (typically via a vector search index) and supplied to the model as context at query time, so answers cite real source material without retraining. Fine-tuning is costly and doesn’t keep content current, temperature changes randomness not grounding, and model size affects cost/latency not accuracy. RAG and Foundry’s grounding tools are central to the refreshed AI-901.

3. Question 3easy

   Which is the BEST description of machine learning?

   • ASoftware that follows only hand-written rules
   • BSystems that learn patterns from data to make predictions or decisions
   • CA type of database
   • DA networking protocol
   Show answer & explanation

   Correct: B

   Machine learning systems learn patterns from data to make predictions/decisions.

4. Question 4easy

   Which responsible-AI principle addresses treating all groups fairly and avoiding harmful bias?

   • AFairness
   • BThroughput
   • CLatency
   • DIndexing
   Show answer & explanation

   Correct: A

   Fairness addresses treating groups fairly and avoiding harmful bias.

5. Question 5easy

   Which responsible-AI principle concerns systems being safe and performing reliably under expected conditions?

   • AReliability and safety
   • BScalability
   • CConcurrency
   • DCaching
   Show answer & explanation

   Correct: A

   Reliability and safety concerns systems performing dependably and safely.