A Generative AI Engineer is creating an LLM system that retrieves news articles from the year 1918 related to a user query and summarizes them. However, the summaries often include explanations of how they were generated, which is undesirable. Which change should the engineer implement to resolve this?