Question 19
Domain 2: Fundamentals of Generative AIA company is using a pre-trained large language model (LLM) to build a chatbot for product recommendations. The company needs the LLM outputs to be short and written in a specific language. Which solution will align the LLM response quality with the company's expectations?
Correct answer: A
Explanation
Adjusting the prompt lets you specify constraints like “keep responses short” and “respond in Spanish,” which directly shapes the model’s output. Prompting is the right control because the LLM is already pre-trained, and the task is to align generation with “the company’s expectations.”
Why each option is right or wrong
A. Adjust the prompt.
Prompting is the only control here that directly constrains generation at inference time for a pre-trained LLM: by specifying “be brief” and the target language in the prompt, you shape the model’s output without retraining. The other evaluation methods in the source material—ROUGE, BLEU, BERTScore, and LLM-as-a-judge—measure output quality after generation; they do not alter the chatbot’s response behavior, so they cannot enforce the company’s required length or language.
B. Choose an LLM of a different size.
Model size changes capacity, not explicit output constraints like length or language.
C. Increase the temperature.
Higher temperature increases randomness and diversity, not concise, controlled responses.
D. Increase the Top K value.
Higher Top K broadens token sampling choices, usually making outputs less constrained.