Ever get frustrated with search engines not understanding what you really want? Researchers are exploring how Large Language Models (LLMs) can ask *clarifying questions* to improve search results. Imagine an AI that asks, "Are you looking for information on Java the programming language, the island, or the coffee?" when you search for "Java." This research introduces AGENT-CQ, a framework that uses LLMs to automatically generate and evaluate these clarifying questions in conversational search. It's a two-step process: first, the LLM generates potential clarifying questions based on your initial query, using techniques like exploring different facets of the topic or varying the "temperature" of the model to generate more diverse questions. Then, another LLM acts as a crowd-sourced evaluator, judging the quality of the generated questions based on metrics like clarity, relevance, and usefulness. Surprisingly, the research found that LLM-generated questions often outperform human-written ones in terms of helpfulness and leading to better search results. One method, called GPT-Temp, was particularly effective, suggesting that tweaking the model's temperature is key to unlocking its ability to ask insightful clarifying questions. While there are still challenges, like replicating the natural flow of human conversation and avoiding overly complex questions, this research shows the potential for LLMs to significantly improve how we search for information by understanding what we *really* mean.