AI-900 Practice Question: Describe features of generative AI workloads on Azure
A company uses a large language model to generate answers to employee questions about internal HR policies. However, the model sometimes produces answers that are factually incorrect or not based on the official policies. To reduce these inaccuracies, the company wants to provide the model with relevant, up-to-date policy documents as extra context before generating a response. Which technique is being applied?
⚠ Common exam trap
A common mix-up: candidates confuse fine-tuning (which modifies the model) with RAG (which augments the prompt with external data), or assume prompt engineering alone can inject new information, when in fact RAG is the specific technique for grounding with external, up-to-date documents.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Grounding with relevant data (RAG)
The technique described is Retrieval-Augmented Generation (RAG), which retrieves relevant, up-to-date policy documents from an external knowledge base and provides them as context to the large language model before generating a response. This grounds the model's output in verified data, reducing factual inaccuracies without modifying the model itself. Option C is correct because RAG directly addresses the need to supply extra context from authoritative sources.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Prompt engineering only
Why it's wrong here
Prompt engineering carefully constructs the instruction or user message, but because it rarely injects external evidence, the model must still rely on its parametric memory for facts. Without reference documents in the prompt, the model can confidently generate plausible-sounding but incorrect answers, especially for specialized or frequently updated policy material. It is a useful technique for steering tone or output format, but it does not provide the factual grounding needed to reduce hallucinations.
- ✗
Fine-tuning the model on policy documents
Why it's wrong here
Fine-tuning updates the model's weights by retraining on policy documents, which changes the model's learned associations rather than supplying per-query context. This process is expensive, requires regular retraining to stay current, and can cause catastrophic forgetting of previously learned information. Unlike grounding, fine-tuning cannot dynamically retrieve the latest policy text for each specific question, so it is less effective for tasks where accuracy depends on specific, up-to-date source documents.
- ✓
Grounding with relevant data (RAG)
Why this is correct
Grounding with relevant data, typically implemented as retrieval-augmented generation (RAG), fetches pertinent chunks from an external knowledge base and inserts them into the prompt at inference time. The model then bases its answer on the retrieved evidence, which drastically reduces hallucination because the content is anchored to known, accessible sources. It also enables traceable citations and immediate updates to the knowledge base without retraining.
- ✗
Using a content filter
Why it's wrong here
A content filter is a safety mechanism that screens model inputs and outputs for prohibited categories like hate, violence, or sexually explicit material. It operates on toxicity and policy compliance, not on factual correctness, so it cannot detect whether a statement accurately reflects a policy document. A false but unbiased and policy-compliant answer would pass the filter unchanged, leaving hallucinations unaddressed.
Go deeper
Related to this question
Learn chapter
Azure AI Language Service
Key term
Model
In IT and AI, a model is a trained mathematical representation that learns patterns from data to make predictions or decisions.
Key term
RAG
Retrieval-Augmented Generation (RAG) is a technique that combines information retrieval with text generation to produce more accurate and contextually relevant answers from large language models.
About these practice questions
Courseiva writes every AI-900 question from scratch — 985 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-900 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-900 exam.