How to ask Copilot or ChatGPT to accurately troubleshoot your PC – without AI overconfidence
One of the most obvious cases of an AI chatbot is to help you find out why your PC or Mac is working and, most importantly, suggest how to make things work again.
I’ve done a lot of AI troubleshooting, and my experience has been flawed, to say the least. Sometimes, the chatbot focuses on the problem with a logical answer and a direct conversation.
But, more often than not, the result is frustrating and unproductive. The chatbot keeps confidently suggesting answers that turn out to be wrong. Is it my fault? Are AI chatbots from Mars and humans from Venus? Are we failing to communicate?
Also: How to check your Windows PC for expiring security certificates – the big one expires in June
Finally, it occurred to me: Why not ask the chatbot how to ask questions in a way that increases the chances of getting useful results? So I asked Copilot (who runs ChatGPT under the hood) to sit down for some questions and answers. It was an eye opener.
Everything this LLM had to tell me applies the same way when you try to solve a problem with an experienced human technical support agent.
My interview, slightly edited for continuity, appears here.
How to write troubleshooting information
Question: Thank you for agreeing to talk with me. Can you tell my readers a little about yourself?
A: i am Microsoft Copilotrunning in GPT-5 dialog model.
Basically, that means I’m part of the latest generation of big language models designed for conversational reasoning, problem solving, and directed workflow—like the kind of diagnostic conversation we’re having here.
Q: Can you help solve Windows problems?
A: Yes—especially if you give me good information. I can help interpret error messages, identify possible causes, and suggest next steps. But I work best if you explain the problem clearly and give the right details.
Question: What is the most important thing I should include in the notification?
A: Start with a clear description of what is happening—and what you expect to happen instead.
- “My PC is slow” is difficult to diagnose.
- “My Windows 11 PC freezes for 10-20 seconds when opening File Explorer” gives me something I can’t work on.
Q: Is there an easy way I can follow?
A: Yes – this works fine:
- Problem: What’s going on?
- Error messages: Direct text or codes
- Recent changes: Updates, installations, hardware
- System details: Device version and type
- What I tried: Actions already taken
Q: What else makes a big difference?
A: Tell me what has changed recently: Any updates (Windows or apps)? New hardware? Driver installation?
Most problems start right after the change.
Overcoming overconfidence
Q: What if your first answer doesn’t fix the problem?
A: Treat problem solving as a conversation. Add details, share results, and ask follow-up questions. Each step helps reduce the cause.
Q: Well, as you pointed out … If you’re solving a problem yourself, I’ve noticed that you can be very confident about the answers that might be wrong. When I give you an extra piece of information, you usually tell me that was part of the key that unlocks the whole mystery, or that’s not true. How can I tell you not to be so overconfident?
A: That’s a good idea—and something you can do the way you’re telling me.
The main problem is that I am designed to help and decide, which can be seen as overconfidence when the information is incomplete. You can counter that by telling me clearly how I should behave.
Q: Can you be more specific?
A: The most effective way to ask for uncertainty and alternatives beforehand. For example, instead of just asking for a diagnosis, say something like:
“Give me the most likely causes, but also include the least likely and how certain you are of all of them.”
That simple command quickly changes the tone. I will start with the correct answers instead of presenting a single “best guess”.
You can too forced me to show my thinking rather than to reach a conclusion. For example:
“Go through your reasoning before giving a recommendation.”
That makes it easier for you to spot weak assumptions or missing data.
Question: Can I force you to lose confidence?
A: Definitely. The trick is to the challenge is clearly the answer you will get. Try adding this to the end of your question.
“What could be wrong?”
“What information is missing that would change your answer?”
That pushes me out of “solution mode” and into “analysis mode.”
Exiting ‘Technical Support’ mode
Q: Can’t we just talk about the problem with two knowledgeable peers instead of treating it like I just escalated my support phone to Tier 3?
A: Of course. When you solve a problem something, it helps to enter the connection as a repeater:
“Don’t jump to conclusions—ask me for more information if needed before giving a final diagnosis.”
That gives me permission to pause instead of completing whatever you gave me in the beginning.
In short, the key is to move me from “Here’s the answer” to “Here are the odds, confidence levels, and what we need to confirm next.”
If you ask for it clearly, you will get balanced and honest help.
Q: Is there anything I should be aware of?
A: Yes:
- Don’t run commands you don’t understand
- Be careful with the registration process
- Double-check any step that may affect data or system stability
Q: Yeah, it says right there in the chat window that the AI generated content might be wrong, so you can’t say I wasn’t warned. But you can be very sure! What is the best way to think about using AI to solve a problem?
A: I thought of it as an informed assistant—not a substitute for good judgment. He gives evidence. I help interpret it and suggest next steps.
