1. Problem

What should I pay attention to when I ask you questions? How can I ask you questions better?

2. Answers from each model

1. Deepseek

To get your questions answered more effectively, follow these structured guidelines:

1. Preparation before asking questions

1) Clarify core goals

• Vague questions: “Help me solve this coding problem”

• Clear question: “How to merge two CSV files using Pandas in Python?”

2) Collect background information

• Example: "I am processing sales data. There are two files with 'order ID' and 'customer information'. I need to merge them by ID. Currently, I get an error when using `pd.merge()`."

2. Tips for asking questions

1) Structural decomposition of complex problems

• Wrong approach: “How to develop an app?”

• Correct Split:

a) Recommend cross-platform development tools (e.g. Flutter vs React Native)

b) Best practices for implementing user authentication

c) How to integrate payment gateway

2) Use terminology accurately

• Error: “My computer crashed, what should I do?”

• Correct: "Windows 10 suddenly blue screens with error code 0x0000003B when running Premiere Pro."

3) Mixed questioning strategy

• Closed + Open combination: "What is the difference between `is` and `==` in Python? (Closed) → In what scenarios are they likely to cause hidden dangers? (Open)

3. Optimization after asking questions

1) Targeted questioning

• If the answer is too general: "Can you give an example of a specific application of a recursive function in traversing a binary tree?"

• If the terminology is not clear: “Can you explain backpropagation in neural networks using an analogy?”

2) Timely feedback and adjustment

• Example: "The `subprocess.Popen` method you recommended earlier solves the problem of starting a process, but now we need to capture the output in real time. How should we modify it?"

4. Notes

1) Balance between simplicity and focus

• Avoid information overload: Provide key parameters (such as code error snippets) rather than full code when describing the context.

• Example: Incorrect question: Paste 200 lines of code → Correct question: Provide the 10 lines of core code that trigger the error and the error message.

2) Polite communication

5. Scenario-based examples

Scenario: Having trouble learning machine learning

• Ineffective questions: “I don’t understand linear regression, what should I do?”

• Effective questioning:

a) Closed question: “What is the specific formula for the loss function in linear regression?”

b) Open extension: “Why is gradient descent more commonly used than analytical solutions in this scenario?”

b) Practical feedback: “I tried the sklearn example but the RMSE value was high. What could be the reason? (with code snippet)”

(2) Kimi

Here are some tips to help you ask me better questions:

1. Clarify the topic of the question

• Before asking a question, first clarify what area you want to know about, such as details about historical events, explanations of scientific principles, or a specific problem in life. This will allow me to more accurately locate the direction of the answer. For example, if you want to know "trade on the ancient Silk Road", after clarifying the topic, I will provide you with a detailed answer from the starting and ending points of the Silk Road, the main trade commodities, trade routes, etc.

2. Specific problem description

• Try to avoid overly general questions. Specific questions are more likely to get accurate answers. For example, instead of just asking "How can I lose weight?", you can ask "I want to lose weight through exercise. I can exercise three times a week for one hour each time. What kind of exercise methods and plans are suitable for me?" This way I can provide you with targeted suggestions based on your conditions.

3. Provide background information

• If the question involves a specific situation or background, it is best to briefly introduce it. For example, if you ask "Why can't I use this software?", you can add "I am using Windows system, the software is the latest version, and it has been working normally before, but it suddenly can't be opened today." In this way, I can analyze the reasons for you from the perspectives of software compatibility, system problems, etc.

4. Ask one question at a time

• If you ask multiple questions at once, your answers may not be detailed enough or the key points may not be highlighted. For example, if you ask "I want to know the principles, applications and future development trends of artificial intelligence", I will try my best to answer, but I may not be able to expand on each point in depth. It is recommended that you split the question into parts, such as asking "What are the principles of artificial intelligence" first, and then ask the next question after I have answered it clearly.

5. Use clear language

• Try to use simple and accurate words to express your problem, and avoid using vague or ambiguous words. For example, "I feel something is wrong with this thing" is not clear enough. You can say "I found that the screen display of this device is flickering. What's going on?"

6. State the expected response format

• If you have specific requirements for the answer, such as a concise summary of the main points, a detailed explanation, or specific steps, you can specify this in the question. For example, "Please give me a brief summary of the main applications of artificial intelligence in the medical field" or "Please give me a detailed description of how to set up the parental control function of the router."

(3) Bean bag

In order to make communication more efficient and obtain more accurate and satisfactory answers, please pay attention to the following when asking questions:

Clear and specific : accurately state the core , such as "Why did Tang Dynasty poetry flourish?" Don't just say "Let's talk about Tang Dynasty poetry." When asking about an event, give the time, location and other elements. When asking about travel tips, clearly state the destination, budget and time arrangement.

Specific details : Complex questions can be broken down into smaller parts. For example , if you are asking about weight loss methods, you can be specific about "how to initially reduce fat within two weeks for sedentary office workers."

Explain background needs : state purpose, report writing, daily understanding, such as "I am writing a paper and need the specific process and impact of the May Fourth Movement."

Ask one topic at a time : Avoid asking multiple unrelated questions at the same time, such as "how to learn English, how to improve writing, how to exercise". Asking separate questions can get more detailed answers.

(IV) General meaning

In order to better ask me questions and get satisfactory answers, you can pay attention to the following aspects:

1. Be clear and specific : Try to make your questions specific and clear. This will help me understand your needs more accurately and give you a more appropriate answer.

2. Provide context : If possible, provide some relevant background information or context. This helps me take all relevant factors into account when answering and provide a more comprehensive answer.

3. Clear instructions : If you have specific needs, such as searching for network information, analyzing link content, or wanting to generate images, please tell me your needs and the expected results directly and clearly.

4. Use the appropriate language style : Choose the appropriate question style based on the type of information you seek. If it is a professional question, use formal language; if it is a daily chat, you can be more casual.

5. Step-by-step questioning : For complex questions, try to break them down into several smaller questions and ask them one by one. This will not only help you gradually understand the problem in depth, but also allow me to provide information more effectively.

6. Feedback and Clarification : If my response is not clear enough or does not fully address your question, do not hesitate to follow up or ask for clarification.

(V) A Word on Literary Mind

When you ask questions, you can pay attention to the following points to better get the answers you want:

1. Clarify the question : Make sure your question is clear and specific. Avoid vague or overly broad questions. Make sure you know the specific information or content you want to know.

2. Be concise : Try to ask questions in short language and avoid long or complex sentence structures. Concise questions are easier to understand and answer.

3. Avoid ambiguity : Make sure your questions do not contain ambiguity or multiple meanings that could lead to misunderstandings or inaccurate responses.

4. Provide background information : If the question involves a specific field or situation, you can provide some background information to help the respondent better understand your question.

5. Check grammar and spelling : Although it is not necessary, a question with correct grammar and spelling will leave a better impression and be easier to understand.

1. Think before you ask : Before asking a question, think about the core of the question and what kind of answer you want.

2. Use search engines : Before asking a question, you can use a search engine to search for relevant keywords to see if someone has already asked a similar question or if you can find some useful information.

3. Refer to sample questions : You can refer to questions I have answered before to learn how to ask clear and specific questions.

4. Be polite : Although it is not a requirement when asking questions, it is always good to be polite and respectful. You can use polite expressions such as "Excuse me" and "Thank you".

5. Follow-up and supplement : If my answer does not fully meet your needs, or you have further questions, you can follow up and ask additional questions in a timely manner.

(VI) Zhipu Qingyan

Here are some things to keep in mind when asking me questions to help you get the information you need:

1. Clarify your question : Try to make your question specific and clear. This will allow me to more accurately understand your needs and provide relevant answers.

2. Be concise and clear : Try to describe the problem in concise language and avoid too much redundant information. This will help me quickly grasp the key to the problem.

3. Ask questions in categories : If the question involves multiple aspects, you can ask questions in steps or categorize the questions. This will help me answer them one by one.

4. Use keywords : When asking questions, using keywords can help me understand your question area more quickly.

5. Avoid ambiguity : Try to avoid using expressions that may lead to ambiguity so that I don’t misunderstand your intentions.

6. Detailed description : If the question is complex, providing some background information or a detailed description can help me better understand the problem.

7. Logic : Please keep your questions logical as this will help me answer them according to your thinking.

8. Polite language : Using polite language can create a good communication atmosphere. Although I am an artificial intelligence, polite communication always makes people feel happy.

9. Check grammar and spelling : While I will do my best to understand what is said, correct grammar and spelling can reduce misunderstandings in communication.

10. Feedback and Correction : If my answer does not meet your needs, you can provide feedback or further correct the question so that I can serve you better.

(VII) Tencent Yuanbao

Here are some tips to help you ask me better questions:

1. Expression

1) Clear and accurate

2) Be concise and clear

2. Provide background information

1) Relevant prerequisites

2) Special Circumstances

3. Identify the type of question

1) Factual issues

2) Questions seeking advice

4. Polite language

1) Use honorifics

5. Avoid ambiguity

1) Check the wording

2) Reasonable sentence structure

3. Commonalities among the Models

- All models emphasize the need to clearly define the scope of the problem and avoid vague statements (such as "code problem" → "Python merge CSV files").

- Requires the use of precise terms or keywords (e.g. "blue screen error code 0x0000003B").

2. Provide background information

- The scenario, tools, and constraints (e.g., "software crashes on Windows" or "only 2 hours of free time per day") where the problem occurs must be described.

3. Structuring and splitting

- It is recommended to break down complex problems into small steps (e.g., “developing an app” can be broken down into tool selection, user authentication, etc.).

4. Keep the language simple

- Avoid redundancy and focus directly on the core (e.g. “Why did my computer freeze?” is better than a lengthy description).

5. Communicate politely

- Most models mentioned using polite expressions such as “please” and “thank you” to improve the friendliness of communication.

4. Differences in the recommendations of each model

I asked Deepseek to summarize the differences between the answers given by each model, as shown below:

5. Complete and effective questioning ideas

Integrate the answers from various models. It is recommended to ask questions according to the following steps:

1. Clarify core goals

- Precise positioning : Use keywords to describe the problem (such as "Python merging CSV files reports an error").

- Avoid being broad : Say no to “How to develop an app?” and instead split the requirements (tool selection → function implementation).

2. Provide background information

- Scenario description : Describe the environment where the problem occurs (e.g., "Windows 10 system, Pandas version 2.0").

- Constraints : Explain the constraints such as time, resources, etc. (e.g. “only 1 hour a day to learn programming”).

3. Structured Splitting Problem

- Ask questions in steps : break down complex questions into sub-questions (e.g. split “develop an app” into “recommend cross-platform tools → implement user authentication”).

- Hybrid strategy : Combine closed (factual) and open (scenario) questions (e.g. “What is the gradient descent formula? → Why is it more commonly used than analytical solutions?”).

4. Language and expression optimization

- Be concise and clear : remove redundant descriptions and ask questions directly (e.g. “Why did my computer blue screen?”).

- Avoid ambiguity : clarify multiple meanings of words (e.g., whether “apple” refers to the fruit or the brand).

5. Attach necessary supporting materials

- Concise information : Provide key code snippets ( 10 lines of core code + error information) instead of complete code.

- Example description : If operation steps are involved, attach the specific methods tried (such as "used `pd.merge()`, error `KeyError`").

6. Communication and Feedback

- Ask polite questions : Use “please”, “thank you” etc. (e.g. “How can I improve my spoken English?”).

- Targeted follow-up questions : If the answer is incomplete, ask for examples or simplification (e.g., "Can you explain backpropagation using an analogy?").

- Adjust in real time : based on answering iterative questions (like "your recommended `subprocess.Popen` works, but how do I capture the output in real time?").

6. Golden rules for asking effective questions

“Precise description (What) + layered refinement (How) + background support (Why)”

- What: Use keywords and terms to clarify the essence of the problem.

- How: Dive into details through split and mixed questioning strategies.

- Why: Provides context and constraints to help respondents understand the root cause of the need.

Example:

"When I was processing sales data in Python, I tried to merge two CSV files (by the 'Order ID' field) using `pd.merge()`, but got an error `KeyError: 'Order ID'`. Both files were confirmed to contain this field and were encoded in UTF-8. What could be the reason? Can you provide troubleshooting steps?" -