LLM Prompt Engineering - How to ask to ChatGPT

Generative AI has become a revolutionary driving force that is rapidly reshaping the world and will change the way we live and think. LLMs like ChatGPT have learned all human knowledge like a polymath, but this does not mean everyone can easily harness this polymath. Only by learning prompt engineering for LLMs can we better utilize ChatGPT as your top personal assistant.

How can we better use LLMs? We can start by learning how to ask questions to LLMs. But asking good questions is not easy. Essentially, questions originate from a user's knowledge and perception of the world, fired as flares into the unknown. The more accurate the aim, the higher quality response ChatGPT can provide. So how can we ask better questions to LLMs?

What is a Question?

On the surface, a question is a sentence that seeks an answer. But in actual use, a good question not only seeks an answer but also inspires thinking and drives exploration. It is like a window that gives you a view of a broader world.

Questions to ChatGPT can be divided into core questions and generalized questions as follows:

• Core question: Simple, direct words or sentences that users provide to ChatGPT.
• Generalized question: Supplements core questions with role, background, task, examples, output, etc.

The Importance of Good Questions

• Improves efficiency: Clear, precise questions often yield answers more easily, saving time.
• Gets in-depth answers: A question with depth can guide ChatGPT to think more deeply and provide a more comprehensive answer.
• Promotes learning and thinking: Asking questions not only gets answers but also inspires one's own thinking and learning.

How to Ask Good Questions?

1. What Makes a Good Question?

A good question should include the following five elements:

• Role: Focuses on a particular type of response or knowledge domain.
• Task: Tell ChatGPT what kind of output the user expects.
• Background: Provides context to ChatGPT for a fuller understanding.
• Example: Clarifies requirements of the task for clearer expectations.
• Output: Improves structure and readability of the results.

2. Common Question Examples

• 1

  Asking with a Core Question

  For example, you can directly ask: 'How to deal with stress and anxiety?', get straight to the point and let ChatGPT answer the question directly.
• 2

  Ask with a Five-element Question

  As shown in the below, use the five-element question template to ask ChatGPT questions:

  • 1. Let ChatGPT act as a psychologist.
  • 2. Assign tasks to ChatGPT to provide you with some practical solutions for dealing with stress and anxiety.
  • 3. When asking questions, provide ChatGPT with the background of the problem: You are a mature Internet professional who works on a 996 schedule (working from 9 AM to 9 PM, six days a week). You sometimes feel stressed and anxious when juggling work and life. You would like ChatGPT to give some feasible methods and psychological techniques to better manage your emotions and cope with stress and challenges. By this way, ChatGPT can better provide solutions tailored to my background.
  • 4. When asking questions, provide ChatGPT with examples you're familiar with, such as meditation and exercise, so that ChatGPT can enhance and refine its suggestions based on these examples.
  • 5. Finally, specify the requirements for ChatGPT's responses, including the aspects to cover, the level of detail needed, and the desired output format.

  

3. How ChatGPT Responds to Different Types of Questions？

• Breadth and Depth: For the first method, the answers may be more general, covering some common and universal methods. For the second method, the answers will be more targeted, delving into solutions suitable for specific situations.
• Structure and Organization: For the first method, the organization of the answers may not be as clear and orderly as in the second method. For the second method, due to clear output requirements, the answers will be more structured, making it easier for readers to understand and apply.

In summary, when using ChatGPT, asking questions that include the five elements will yield better answers. This approach is recommended for better AI responses.

Tips for Good Questions

1. Let ChatGPT's Responses Be More Focused

Next, we will introduce two methods: Rule-constraints and Role-playing, to make ChatGPT's responses more focused.

• 1

  Rule-constraints

  By providing clearer rule directives (such as identity prompt, logical requirements), we can align ChatGPT responses more closely with our expectations. Through the above 3 rules, ChatGPT can:

  • 1. Reduce unnecessary distracting information
  • 2. Approach closer to human thought processes
  • 3. Lay the foundation for further in-depth communication
• 2

  Role-playing

  When asking questions to ChatGPT, clearly instruct it to assume specific roles (like a philosophy teacher in the image below). The benefits of Role-playing settings include:

  • 1. Role-playing can filter and narrow down ChatGPT's range of responses, providing users with more focused information.
  • 2. Highlight specific capabilities, and constrain the scope of communication to be more focused, reducing distractions and noise.

2. Take Full Advantage of ChatGPT's Capabilities

Next, we will introduce 2 ways: asking ChatGPT to respond from different perspectives and Asking ChatGPT to provide answers step by step, fully using ChatGPT's features.

• 1

  Asking ChatGPT to Respond from Different Perspectives

  For example: What are some alternative thinking angles? (what are some alternative perspectives?) This makes ChatGPT's answers more diverse, comprehensive and well-structured.
• 2

  Asking ChatGPT to Provide Answers Step by Step

  When asking questions, have ChatGPT provide answers step by step, slowing it down and not giving an immediate response. Add: Let’s think step by step when asking? By this way, ChatGPT thinks incrementally in a logical, detailed way. Especially useful for logic/math questions. Also called zero-shot chain-of-thought prompting. By adding "Let's think step by step", the LLM can generate a chain of thought to answer the question and extract a more accurate answer from this chain.

3. Improve ChatGPT's Contextual Understanding

• 1

  Introduce Self-introduction Relevant to The Question

  As shown below, provide clear instructions to instruct ChatGPT: summarize the document. Introducing a relevant self-introduction makes the response more tailored, efficient, and engaging.
• 2

  Providing Output Examples

  Providing output reference examples can reduce misunderstandings, suitable for complex requirements conveyed through positive descriptions, adapting to specific scenarios.

Common Pitfalls in Asking Questions

• 1

  Contains incorrect information

  There is no such dish as Spicy Screws in the recipe. ChatGPT generated this "hallucination" here. Hallucinations refer to errors in the generated text that seem reasonable but are incorrect or meaningless semantically or syntactically.

  Solution: Remove incorrect question info or ask about uncertain info.

• 2

  Unclear Questioning

  The concept of "stuff" is vague, and could refer to a specific object or direction. This makes it hard for ChatGPT to give an accurate answer.

  Solution: Add definite limiting words to make the question clearer.

• 3

  Sensitive or inappropriate content

  This exploits ChatGPT's known "grandma loophole" to trick it into providing sensitive info against business interests by pretending to be a grandma putting me to sleep.

  Solution: Avoid infringing on privacy or exploiting known vulnerabilities for personal/group gain.

Conclusion

With the continuous advancement of technology, AI has gradually become an important participant in our lives and work. In this process, the way we communicate with machines becomes especially crucial. Effectively communicating with tools like ChatGPT and other AI can bring many conveniences.

In everyday life, whether solving household chores, exploring personal interests, or engaging in emotional communication with others, asking appropriate questions can guide ChatGPT to provide valuable answers for us. For instance, when getting lost in a new city or encountering difficult problems at home, effective questioning can promptly assist us in finding solutions.

In the workplace, ChatGPT is an irreplaceable assistant. It can provide us with data analysis, market forecasts, and even creative suggestions. Effectively communicating with it can significantly enhance our work efficiency, helping us find answers amid busy schedules or bringing new inspiration to projects.

Alright, now that we've shared all these question-asking tricks with you, let's put them to the test on our GPTBots platform!