Question

A product manager notices that an LLM-based assistant sometimes invents feature details when answering employee questions about the company's internal software. She wants to keep using the same model but improve answer accuracy. Which prompting approach most directly addresses this problem?

Accepted Answer

Include the relevant internal documentation as context in the prompt so the model grounds its answers in that source material

Answer

Increase the model's temperature setting so it generates more varied and creative responses

Answer

Ask the model to answer as quickly as possible using fewer tokens to save on cost

Answer

Rephrase every question as a single word to reduce the size of the input prompt