Whether you’re a tech boffin or prefer to keep things analogue, you’ve likely heard of ChatGPT. It’s the world’s most popular AI chatbot, boasting hundreds of millions of active users per week.

This popularity isn’t without good reason. From writing poems and answering questions to generating custom images and videos, ChatGPT can be useful in various aspects of life, and its user-friendly design means that anyone can navigate it with ease.

However, there are still some tricks for getting the most from ChatGPT – and it all starts with ‘prompts’, which are the text or data you enter into the chatbot to generate responses. The content and structure of these inputs can have a huge impact on the answer you get, so it’s worth thinking about how to make them as effective as possible. In the tech world, this is called ‘prompt engineering’.

So, for anyone just starting out with ChatGPT, here are nine tips to get better results.

Key takeaways:​

  • Get specific – the more useful details you include (within reason), the more helpful your responses are likely to be.
  • Use multiple prompts – review the responses ChatGPT gives you and improve with further prompting.
  • Experiment with other AI chatbots – remember, ChatGPT isn’t the only tool out there.
  • Check, check, check facts – ChatGPT isn’t perfect, so always check responses, especially when using it for information.

1. Ask for multiple answers

One of the easiest ways to get better responses from ChatGPT is to ask for more than one answer. For example, if you need help writing an email to request a product refund, you could ask for three drafts, each varying in firmness.

Variety means we’re more likely to get an answer we like. You might find that you prefer one response entirely over the others, or take parts of each response to build one that works for you. In my experience, I’ve found the latter is usually the case, which is another reminder why a single answer usually isn’t enough.

Asking for multiple answers also prevents us from becoming disconnected and relying on AI to do all the work for us. As we decide what we like about each answer and why, we’re actively engaging with the task at hand, which often leads to better results. Remember: AI is best used as a partner for human creativity, not a replacement for it.

2. Include plenty of context

As with talking to a human, the more context you provide ChatGPT with, the better the answers will be. Details about your situation can help the AI chatbot ‘understand’ the nuances of your question and better align its response to your needs.

Say you asked ChatGPT: “How do I deal with a difficult colleague?” It may offer some helpful points, but much of the advice will be vague and not necessarily applicable to your specific situation.

Instead, you might find it more helpful to try something like this…

“I manage a small sales team of six people. Most of the team gets along well, but one member frequently interrupts others in meetings and disrespectfully dismisses their ideas. How can I address this situation? I want to avoid creating conflict, but I also want this employee to take what I’m saying seriously. What are some practical ways to handle this one-to-one conversation?”

Not only was the advice generated by this prompt more practical and actionable, but it also offered helpful examples that applied directly to the situation described above.

Include plenty of context

3. Assign ChatGPT an identity

Many experts say that giving ChatGPT a specific identity can also improve the quality of its responses. This could be a profession (such as a teacher or fiction editor) or a more general role in your interaction (like a friend or job interviewer). It’s up to you to choose what’s best based on your needs.

This easy trick helps give the AI chatbot a specific point of view in your conversation, changing its priorities and the way it views your prompt, and making its answers more useful.

For example, say you’ve written a cover letter for a job application, and you’d like some suggestions on how to improve it (remember to remove any sensitive information and personal details you don’t want it to remember). Telling it to be a writing coach will give you more advice on things like structure and tone, while asking it to be a hiring manager will prompt it to focus more on the content.

To action this tip, just add a line like “Act as a [insert role]” to the beginning of your prompt.

4. Give ChatGPT a target audience

Much like the previous tip, giving ChatGPT a target audience can also help you get more specific, helpful responses. For example, if you want help planning a night in with your grandchildren, you could say something like: “Give me examples for fun things to do on a rainy evening with children aged 8-11.”

However, the target audience doesn’t have to be other people; it could also be you. When using ChatGPT to learn about new topics, some people find it helpful to say something like: “Explain X as if I know nothing about the subject”. This will ensure it gives you a basic understanding to begin with that you can build on with further prompts.

This approach also works the other way around. If you’re already somewhat knowledgeable about a subject, let ChatGPT know, and it’ll answer with this in mind. For example, if you’re an experienced potter looking for ways to take your ceramic creations to the next level, it can give you suggestions based on your skill level, using terminology that you’re already familiar with, rather than wasting time explaining the basics.

Give ChatGPT a target audience

5. Refine responses with further prompts

The previous tips have all been about writing the best possible first prompt. While this can save time, it’s worth remembering that you don’t have to get everything right from the start. And sometimes, writing a few short, separate prompts instead of one long one can make the AI less likely to misinterpret your meaning and lead to better results.

So why not try breaking your task down into steps and refining responses with further prompts?

Here’s a simple example…

Prompt 1: “Suggest some cheap and easy meals to make on weeknights.”

[review response]

Prompt 2: “Make sure they are all dairy-free, don’t include red meat, and use as many store cupboard staples as possible.”

[review response]

Prompt 3: “Organise these into a calendar schedule with the quickest two recipes on Tuesday and Wednesday. And write me a shopping list including everything I’ll need for the week.”

In the AI world, this technique is called ‘iterative refinement’. It essentially means beginning with a broad, baseline task and slowly asking it to adjust the content, tone, or structure based on how useful you find the previous responses. Think of it as slowly editing a piece of writing through successive drafts.

6. Provide example answers

Sometimes, ChatGPT can be a little frustrating. You can do all the prompt refinement in the world and still not get a response you’re happy with. When this happens, you could try providing some example answers.

Say you’re using ChatGPT to help you write copy for your business’s website, but no matter what you try, it doesn’t sound quite right. In this case, you might find it useful to provide the chatbot with similar passages you’ve written before, or even snippets from other websites that you’d like yours to sound like.

Often, providing examples is a much more effective way to communicate what we’re looking for than just trying to explain it. It’s especially effective for style and tone, but it can also help with things like formatting, too, if you need the response to be structured in a certain way.

7. Ask ChatGPT for prompt advice

Another unique tip that some experts suggest is asking ChatGPT for advice on how to prompt it more effectively. For example, if you’re seeking advice on how to tackle the slug problem in your garden, you could try something like this…

Example prompt: “I want to stop slugs eating the plants in my garden. What information do you need to help me to achieve this?”

In response, ChatGPT asked me for follow-up information on things like my garden layout, the plants present, and where I lived. It also requested more insight into how bad the slug problem was and my preferences for control methods – i.e. organic/natural, chemical, or physical barriers.

As someone who knows very little about gardening (and slugs), I found this approach very helpful. I wasn’t aware that I should be considering many of these factors in my original prompt, and once I answered with all the relevant information, ChatGPT gave me a tailored plan on how to tackle my infestation.

Ask ChatGPT for prompt advice

8. Try other AI chatbots

ChatGPT might be the most famous AI chatbot on the market, but it’s far from alone – and each one has different ways of answering prompts. So, if you’re looking for better results, it’s worth trying others, too.

For example, if you’re looking for underrated walking holiday destinations throughout the UK, give ChatGPT a go first. But then try entering the same prompt into other tools and see how the responses differ. Some of the most popular AI chatbots include Gemini, Perplexity, Claude, and Copilot.

It’s helpful to think of this like gathering multiple perspectives. Remember: in real life, if you want to find out the best way to do something, asking three people is usually better than asking one – and the same goes for AI chatbots.

Different AI chatbots also have different task-specific strengths and weaknesses. So, depending on what you’re using them for, choosing your tool carefully can make a big difference. For example, Claude is great for creative writing, while Perplexity excels at web searching.

This article from PC Mag offers a brief idea of which AI chatbots are good for different tasks, but the best way to find out is to experiment with them yourself.

9. Always double-check responses

Perhaps the most essential thing to remember when using ChatGPT and similar tools is to check their responses carefully. Not only can AI chatbots misinterpret the meaning of your prompts, but they can also respond with incorrect information. The reason for this is to do with how they work.

Because AI responses are often very articulate, it’s tempting to mistake what they’re doing for ‘thinking’. But it’s important to remember that they aren’t.

Instead, they’re trained on huge amounts of data and use this to learn patterns in language. They produce responses by predicting what words or phrases most likely come after others. Because there’s no understanding involved, this method can lead to mistakes.

For example, AI chatbots can often provide answers that, while sounding confident, are actually incorrect or nonsensical. These are called ‘hallucinations’. If you’ve ever used ChatGPT for recipes, you might’ve noticed some wild suggestions when it comes to measurements (like 400g of sugar instead of 40g). Mistakes can also result from incorrect data being used to train the AI.

With this in mind, we recommend using these tools for inspiration rather than information. Though if you want to learn about a new topic with an AI chatbot, remember to fact-check everything it says against trusted sources.

Final thought…

The tips above should help take your ChatGPT experience to the next level – but it’s also important to take our advice with a pinch of salt and experiment for yourself.

AI is a constantly evolving field, and the companies behind tools like ChatGPT are updating them at breakneck speeds. So what works well today might not have the same result tomorrow.

For more help getting started with AI chatbots, check out our article: Everyday AI – 11 fun and helpful ways to use ChatGPT.

Or, you might like to join our free 60-minute Simple Guide to AI on Your Phone over on Rest Less Events. This session covers everything from tweaking emails to identifying household objects.

Do you want to see more AI-related content from Rest Less? If so, what would you like us to cover? We’d love to hear from you in the comments below.