AIF-C01 Practice Question: A data scientist is evaluating whether to use…

A data scientist is evaluating whether to use fine-tuning or Retrieval-Augmented Generation (RAG) for a legal document analysis application. Which TWO statements correctly describe when to use each approach?

⚠ Common exam trap

AWS often tests the misconception that fine-tuning can inject new factual knowledge or enable real-time updates, when in fact RAG is the appropriate technique for dynamic or external knowledge integration.

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

✓

Use fine-tuning to teach the model a specific output format or style

Option B is correct because fine-tuning adjusts the model's weights on labeled examples, which is the right technique for teaching a consistent output format, tone, or style (e.g., always returning structured legal citations in a fixed schema). Option E is correct because RAG retrieves documents from an external knowledge base at inference time, so when the underlying corpus changes frequently, you simply update the index/vector store rather than retraining the model. Option A is wrong because RAG reduces hallucinations by grounding responses in retrieved external data, not by working 'without any external data.' Option C is wrong because real-time information access is achieved through retrieval (RAG) or tool calls, not fine-tuning, which bakes in static weights at training time. Option D is wrong because fine-tuning is unreliable for injecting new factual knowledge; RAG is the preferred approach for adding or updating facts.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Use RAG to reduce model hallucinations without any external data
Why it's wrong here
RAG retrieves from an external knowledge base; without any external data there is nothing to retrieve, so it cannot ground responses or reduce hallucinations. It is tempting because RAG is the standard grounding technique, and it would be correct when current or proprietary documents must be supplied to the model at inference time.
✓
Use fine-tuning to teach the model a specific output format or style
Why this is correct
Fine-tuning adjusts model weights on labelled examples, embedding a consistent output structure or tone into the model itself. This suits the legal application when a fixed response format or style is required, rather than retrieving external document content at inference time.
✗
Use fine-tuning to enable the model to access real-time information
Why it's wrong here
Fine-tuning bakes patterns into weights at training time and cannot retrieve live data, so it cannot supply real-time information. It is tempting because fine-tuning does update a model, but real-time access is exactly what RAG's external retrieval provides.
✗
Use fine-tuning to inject new factual knowledge into the model
Why it's wrong here
Fine-tuning adjusts behaviour and style rather than reliably storing new facts, and it risks hallucination on unseen knowledge. It is tempting because training on documents appears to teach facts, yet RAG's retrieval is the mechanism for injecting current factual knowledge.
✓
Use RAG when the knowledge base changes frequently
Why this is correct
RAG retrieves documents at inference time from an external store, so updating the knowledge base requires no retraining. This satisfies the frequently-changing knowledge constraint: new or amended legal content is indexed and immediately available, unlike fine-tuning, which bakes knowledge into weights.

Go deeper

AIF-C01 Practice Question: A data scientist is evaluating whether to use…

Why each option matters

Option-by-option breakdown

Related to this question