Techniques to Improve Generative AI Model Output →mediumMultiple Select

Generative AI Leader Practice Question: Techniques to Improve Generative AI Model Output

Which TWO techniques are most effective for improving the quality of a generative AI model's output when summarizing complex documents?

⚠ Common exam trap

A common misconception is that increasing output length or adjusting sampling parameters like top_p and temperature universally improves output quality, when in fact these parameters must be tuned carefully for the specific task and can degrade summary quality if misapplied.

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

✓

Providing few-shot examples of ideal summaries

Option A is correct because few-shot prompting supplies the model with concrete examples of the desired summary format, style, and level of detail, which conditions the model to reproduce that structure and improves fidelity when summarizing complex documents. Option B is correct because a larger, more capable model such as PaLM 2 has greater capacity to comprehend long, intricate documents and generate more accurate, coherent summaries than a smaller predecessor like PaLM. Option C is not appropriate because increasing max output length only allows longer responses; it does not improve summary quality and can even encourage verbosity rather than conciseness. Option D is not appropriate because setting top_p to 0.1 sharply narrows nucleus sampling, reducing diversity and making the output more rigid without addressing summarization quality. Option E is not appropriate because raising temperature to 0.8 increases randomness, which tends to introduce inaccuracies and inconsistency rather than improve the quality of document summaries.

Answer analysis

Option-by-option breakdown

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

✓
Providing few-shot examples of ideal summaries
Why this is correct
Few-shot examples steer the model by conditioning it on concrete input-output pairs, so it infers the desired summary structure, length and tone directly from the prompt. This satisfies the stem's demand for higher output quality on complex documents without retraining, unlike prompt-only instructions that leave format ambiguous.
✓
Using a larger, more capable model (e.g., PaLM 2 instead of PaLM)
Why this is correct
Scaling to a more capable model such as PaLM 2 raises the parameter count and training-data breadth, improving the model's inherent language comprehension and reasoning. This directly addresses the stem's constraint of summarising complex documents, where deeper semantic understanding is required to condense dense, technical material accurately.
✗
Increasing max output length significantly
Why it's wrong here
Extending max output length only permits longer responses; it does not reduce hallucination or improve fidelity when condensing complex documents, and invites padding. This setting suits tasks needing long-form output, such as report generation, rather than summarisation quality.
✗
Setting top_p to 0.1
Why it's wrong here
Setting top_p to 0.1 restricts sampling to a narrow token nucleus, which reduces diversity and can omit relevant content when summarising complex documents. Top_p tuning suits tasks needing deterministic, factual short answers, not comprehensive multi-faceted summaries.
✗
Adjusting temperature to 0.8
Why it's wrong here
Raising temperature to 0.8 increases sampling randomness, which introduces variability and factual drift into summaries where fidelity matters. Temperature tuning suits creative drafting tasks such as brainstorming or marketing copy, not document summarisation, where low values and grounded prompting techniques are required.

Go deeper

Generative AI Leader Practice Question: Techniques to Improve Generative AI Model Output

Why each option matters

Option-by-option breakdown

Related to this question