How to get better answers from AI chatbots (prompting basics)
The quality of answers you get from an AI chatbot depends on the quality of the questions you ask. It isn’t because the AI is being picky; it’s just how these tools work. A vague question gets a vague answer. A specific question gets a focused one. Getting better results doesn’t require any special knowledge. It’s just about being more thoughtful about how you communicate.
Be specific instead of vague
The most important rule is this: give your AI as much context as you reasonably can. Instead of asking “How do I fix my email?” tell it which email service you use, what the problem is, and what you’ve already tried. Instead of “What should I cook?” say “I have chicken, peppers, and onions. I have 20 minutes and I don’t like spicy food.”
Specificity matters because AI chatbots can only work with the information you give them. They aren’t mind readers. The more detail you provide upfront, the less time you spend clarifying what you meant.
Think of it like asking a human friend for advice. “I’m confused about my job” is different from “My boss keeps assigning me projects that aren’t in my job description, and I’m worried about saying no because I’m only six months in.” The second version gets you better advice because the person understands what’s actually happening.
Give the AI a role or angle
Sometimes it helps to tell the AI what perspective you want it to take. You might say, “Act as a beginner-friendly writing teacher and explain how to write a strong paragraph,” or “I’m a parent new to this technology. Explain password managers like I have no technical background.”
This doesn’t require magic phrases. You’re just helping the AI understand who it’s talking to and what would be useful for that person. It naturally shapes the explanation to match your needs.
Ask for the format you want
Tell the AI how you want the answer. Do you want a numbered list? A paragraph? A pros-and-cons table? A script you can use? Do you want it short enough to fit in an email, or as detailed as possible?
For example: “Give me a one-paragraph summary,” “List the five most important things to know,” or “Write this as a professional email I can send to my landlord.” This prevents you from getting back something that’s the wrong shape for what you actually need.
Ask for step-by-step explanations
If you’re learning something new, ask the AI to break it into steps. “Walk me through how to set up a password manager” or “What are the steps to back up my phone?” will often get you a clearer, more actionable answer than a general explanation.
You can also ask for explanations at a particular level of detail. “Explain cryptocurrency like I’m ten years old” forces the AI to use simpler language. You can also ask it to “assume I know nothing about photography” or “assume I understand spreadsheets already.”
Build on responses that are close but not quite right
If the AI gives you something that’s heading in the right direction but misses the mark, just tell it what’s wrong and ask it to adjust. You don’t need to start over.
For example: “That’s helpful, but I need this to work offline” or “You’re on the right track, but that’s more formal than I need.” The AI will adjust based on your feedback. This back-and-forth is one of the most powerful ways to use these tools. You’re refining your request in real time based on what you’re getting back.
Share examples of what you want
If you’re asking for something creative or subjective—like a tone of voice, a writing style, or an approach to a problem—showing an example helps a lot. You might paste in a paragraph that’s written in the tone you like and say, “Write something like this, but about X instead.” Or show an example of a budget template you like and ask the AI to create something similar for your needs.
Avoid unnecessary politeness
There’s no need to say “please” and “thank you” to an AI. It doesn’t matter to the AI, and it just takes up space in your prompt. Be direct. “Summarize this article in three sentences” is better than “Would you mind possibly summarizing this article for me?”
Know when the AI is out of its depth
Even well-asked questions sometimes get weak answers if the topic is outside the AI’s training or requires real-world knowledge the AI doesn’t have. If you ask about very recent events, changes that happened after the AI’s training data stops, or something that requires hands-on experience, the AI might not know.
If you’re asking about something that requires current information (like today’s weather or current prices), mention that limitation in your question and ask the AI to help you think through the problem anyway.
Clear communication is key
Getting better answers from AI comes down to clear communication. The clearer you are about what you’re asking, what context matters, what format you want, and what your constraints are, the more useful the answers will be. None of this requires technical knowledge. It’s just about being specific, iterating when needed, and thinking like you’re explaining something to a helpful person who’s missing some context.
Further reading: OpenAI’s Tips for getting better results and Anthropic’s Prompt engineering guide offer deeper technical guidance if you’re curious about the underlying principles.