What is Question Refinement?
Question refinement is a technique used to improve the quality and relevance of answers received from a Large Language Model (LLM) like ChatGPT. The idea is to start with a general or broad question and then gradually narrow it down by asking more specific follow-up questions. This iterative approach allows you to hone in on the exact information you’re looking for.
Why Use Question Refinement?
LLMs are designed to provide information based on the queries they receive, but they are not always perfect at discerning the user’s intent from a single, ambiguous question. By refining your questions, you make it easier for the LLM to understand what you’re asking, which in turn increases the likelihood of receiving a more accurate and useful answer.
Examples of Question Refinement
Let’s say you’re interested in learning about renewable energy. You could start with a broad question like, “Tell me about renewable energy.” The LLM might provide a general overview, covering various types like solar, wind, and hydro energy. If you’re specifically interested in solar energy, your next question could be, “What are the advantages of solar energy?” The LLM will then focus on that particular aspect, giving you a more targeted answer.
Here’s how the refinement could look:
- General Question: Tell me about renewable energy.
- Possible Output: Overview of renewable energy types like solar, wind, and hydro.
- Refined Question: What are the advantages of solar energy?
- Possible Output: List of benefits such as sustainability, low operating cost, etc.
Further Refinement and Outputs
You can continue to refine your question based on the information you receive. If one of the advantages mentioned is “low operating cost,” and you want to know more about that, your next question could be, “Why is the operating cost of solar energy low?”
Possible outputs at this stage might include details about the absence of fuel costs, low maintenance requirements, and government incentives that make solar energy economical in the long run.
Question Refinement Pattern
The question refinement pattern means giving ChatGPT a prompt that asks it to suggest an improved question whenever you ask it something on a given subject. For example: “From now on, whenever I ask a question, suggest a better version of the question to use instead”. A more specific example might be: “From now on, when I ask a question about gaming, suggest a better version of the question that focuses more on non-gaming activities, and ask me if I would like to use it”: