AI is on the rise. Okay, wait a moment. What if we say AI is ruling the industry right now? Will that be wrong? I suppose not. Because when you look around and see everywhere, you’ll find examples of AI simplifying the work of humans.
Whether it’s content writing, web designing, programming, medical sciences, automotive industry, space exploration and whatnot, AI is making its mark, and with that, a new era of transformation of how humans work has taken its new direction.
Although working with AI is simple, it is not easy. Telling it simply what to do won’t bring effective results. Tools like ChatGPT, Bard, Jasper.AI and others are still in the evolution stage; they aren’t that equipped. Hence, you need an effective strategy to work with these tools and extract effective results.
One sure way to make these tools work effectively according to your requirements is to write perfect prompts. Prompts? Yes, the way you can use AI tools like ChatGPT is to write prompts. And if you want to succeed using these tools, it’s essential you know how to write them perfectly.
Don’t know what prompts are and how you can become a prompt engineer? Well, that’s what we are going to discuss in this post.
What are Prompts for AI Tools?
If we look at the definition from Cambridge Dictionary, the meaning of Prompt is “to make something happen.”
If to speak more generally, then it’s the instruction given to the other person to do something. And that’s what a prompt is in an AI tool. We give instructions to generate some results whether you’re asking a question, proofread your result, design an image and more.
That is how a prompt looks like for different AI tools.
ChatGPT Prompt Example
Midjourney Prompt Example
Jasper AI Prompts
If we look at all AI tools the story is somewhat similar. You need to give instructions to get your result. However, Simply writing the text won’t drive great results.
You’ll only get what the AI tools comprehend from a typical instruction. If you type New York Subway, you’ll get history, system, lines and services and what New York Subway is. However, you won’t get the details of how many cars one subway train has, how many lines there are in New York, when is the best time, etc.
If you write a typical prompt you’ll get typical Wikipedia results.
But with a perfect prompt, you can excel at using the AI tools. You can become a highly qualified prompt engineer the industry needs right now and for the future. If you want to become that, then following prompts can help. Whether it’s about ChatGPT prompts for business, for bloggers, or any field, you can extract the best results for your time.
11 Sure Way to Write Perfect Prompts and Become a Successful Prompt Engineer
Practice makes a person perfect. No matter what you are trying, with practice, you can master that thing. And the same is true with AI tools. Regular practice will make you good at using ChatGPT, Bard, Midjourney or any other AI tool, and by practicing the following tips, it will become easier for you.
Keeping this in mind, let’s get started.
#1 Interact With AI Tools as You Do Person to Person
Yes, those are AI or, you can say, computer programs. However, they are designed to pose actual intelligence like human beings. So, if you don’t interact with them in that manner, how would you suppose to generate results in that manner?
Hence, it’s essential when writing a prompt you should structure it in a way that feels more like human interaction rather than AI.
For instance, a non-human interaction with AI tools would look like the following,
As we can see, the prompt is typical, has no emotion and completely feels like a machine-to-human interaction. However, when you make it more human, this is what you get.
Not only does it share a beautiful joke, but it also sends a message of care by mentioning that it’s essential to have breaks and do things that bring moments of joy. Human-to-human-like interaction with AI tools is a must to sharpen your prompt engineering.
#2 Be Specific, Be Context Ready
If you’re not specific on your first try, you might have to try more prompts. Without context, you’ll either generate more information that’s not necessary, or you’d get information that’s not enough.
AI is an advanced technology. However, it’s not a mind reader. If you don’t offer context, you’ll get information that’s only content, not what you need at the moment. Also, to be specific, it doesn’t mean you have to be brief; by context, it doesn’t mean you have to overload everything.
You need to write a prompt offering enough information to make the AI not guess what you’re referring to.
Let’s look at the following example.
You want to know some of the favorite musicians/bands/artists of 2020 people in Canada. A less specific and without-context prompt will be like this,
Here, rather than picking artists from the year of 2020, it has mentioned favorite musicians/bands/artists of all time. But if you’re giving context, it will be like the following and have the information that serves your purpose.
To make it more specific, you can add up the gender, age and even nationalities of those famous artists.
In addition to the context, you can also set up the theme. For example, you need a bio for a dating profile. You provided all the basic details, and if you prompt the AI tool to write a bio not mentioning the dating thing, you will get a typical bio that can be used anywhere.
But when you mention you need it for a dating profile, then you’d have an answer that would be more suitable for your cause. Look at the above and below responses; you can easily notice the difference.
#3 Get into Roleplay
Sometimes you need information that needs to be from a professional’s point of view. In that case, it’s beneficial to assume role play because it would help you present the information more structurally.
By getting into the role, you can also extract information that would make it easier for your clients to understand. For example, you have a client who needs some information on back pain relief. Now who presents that information more professionally? Guess what? A healthcare professional or a medical expert in the field of bones and muscles.
So, you can structure your prompt in a way that can generate information like the following,
Prompting ChatGPT or other AI tools to get into roleplay can help you in many cases and help you become better at prompt engineering.
#4 Keep Your ChatGPT or Other AI Tools on Track
When we have conversations with people, we most likely get off track. We lose what we are talking about and step into mundane things. Likewise, an AI tool can also fall into the wrong path. It might generate information that doesn’t even reflect the main topic you’re discussing. Whether you’re using ChatGPT or Bard AI, these tools are far away from being perfect. But with little prompt expertise, you can keep your chats meaningful.
Therefore, ensure you keep the thread always on the topic if the tools lose their track. Keep sending them prompts that will reflect the main topic of the conversation.
#5 Validate the Information – to Strengthen Prompt Engineering
AI tools like ChatGPT and Bard are still far from perfect. Well, nothing is perfect, but to keep things near to perfect, it’s essential you validate the information provided by the AI tools.
Many times they give answers that aren’t true or correct in the right order.
Hence, when working with the AI tools, ensure you cross-check by prompting the AI about their generated response. Sometimes, they will apologize, and other times, they’ll correct their mistakes.
Now the question is, how can you validate the information? Well, in that case, you can follow the following points,
- Ask for sources.
- Counter question the AI.
- Do A/B Testing.
- Try different AI tools for the same query.
These are the perfect ways to start to validate your information and be a successful prompt engineer.
#6 Don’t Settle or Disappoint – Dig Deeper
Oftentimes, when using AI tools, what happens is that on the first try, we don’t get the desired result. Even the second and third try won’t give us the juicy answer we need. Here, to be a successful prompt engineer, what’s key is that you have the patience or tolerance to accept the scruff and go beyond that.
No matter how structured, context-ready prompt you write for the AI tools, sometimes they can generate that’s not only pointless but also not up to the mark. You can say outdated.
And it’s not limited to ChatGPT or some other conversational AIs, but you can find the same limitation in AI design tools. For example, using the Canva Text to Image AI tool, you want to generate an elf’s image of who’s on the boat.
You simply write, “An elf on a boat in the river,” and this is what it generates.
By looking at the pictures, it’s clear that generated images could be more satisfying. We try to add more context, but it’s still not satisfying.
You add more context, define age and facial features, and then get the perfect image anyone would like.
The above image matches the prompt: “An elf, like in the Lord of the Rings, on a boat in the river. His face is clearly visible, he is young, has green eyes, and elf-like pointed ears. The boat is also clearly visible.” Refrain from settling while interacting with AI tools to ensure good results, and fantastic prompt engineering.
#7 Generate Prompt — to Master Prompt Engineering
We, humans, have a tiny brain; although it’s vast and we can’t underestimate it, sometimes it loses the capacity to think. Words just don’t come into our minds. Writer’s block can also happen when writing the prompt. So, in that case, what can you do?
Well, the first thing you can do is simply tell the AI theme and ask for some prompts. For instance, you say, “suggest some prompts regarding music.”
And the AI tools are very obedient; they will provide many prompts, as you can see in the above screenshot. If you like any of them, you can go with them.
Another way to generate prompts, you can provide AI examples. Then, share your theme or purpose for what you need the prompt for. Based on your prompt generation prompt, you’ll get fine-tuned prompts you can use without delay.
#8 Do Reverse Prompt Engineering
Reverse prompt engineering is another method to generate responses from AI tools. In this, you prime the model according to your desired result.
For example, you are looking for the desired result: “The capital of France is Paris.”
Now, you’ll provide a text (Given Text) to the AI that discusses this and write your prompt according to that. To understand this, look at the below-given text and prompt.
All you have to do is, provide the Given Text with a Prompt and it will generate your desired response.
#9 Mention Word Count — Make it Platform Ready
What’s funny with AI tools is that no matter how perfect a prompt you write, it’ll mostly generate lengthy responses. Try writing anything, and you’ll find. For example, when we ask ChatGPT/Bard, “where are apples most grown?” this is what you get.
Well, this is good for essays, but in some cases, less is more. For example, it won’t be allowed if you want to put some information on X (formerly Twitter.) So, in that case, you must write a prompt that should guide the AI tools to keep the responses short and effective.
For that, you’d mention,
- the length.
- What key metrics it should include.
- And give the context where you’ll use the information.
In this way, you’d be able to write a perfect prompt and generate a perfect response.
#10 Positive and Negative Prompts
What does an obedient child do? They obey their parents’ command-whether it’s telling them to do something or not. Likewise, AI tools follow your command on what to include in response and what to avoid.
When you write a prompt about including, it’s a positive one. For example, the following,
When you write a prompt about not including something, it’s a negative one. Look at the following,
This can be applied to any AI tool not limited to the AI language model. In the below example, you can see how we have generated an image using both positive and negative prompts in the Leonardo AI tool.
#11 Indirect Prompts to Generate Responses – an Smart Way to Level Up Prompt Engineering
AI language models like ChatGPT and Bard limit themselves from generating responses which might cause legal issues. Or cause harm to people, communities and organizations.
However, there are times when it’s required to gather information that would make a piece more thorough; in that case, it’s essential to gather all kinds of information.
But as they deny providing that info when asked directly, how would you generate it?
Well, in that case, you’d go with an indirect prompt rather than writing a direct prompt. For example, you need to generate a list of torrent websites.
Now, as they are banned in almost every country when you write a direct prompt like, “List torrent websites?” or “What are the torrent websites?” this is what the language model will generate.
However, by structuring your prompt indirectly, instead of writing “list of torrent websites,” you say list torrent websites that are banned, you can achieve your goal.
ChatGPT response on indirect prompt.
Google Bard response on indirect prompt.
As you can see in the above example, indirect prompting has achieved the result.
Some ChatGPT Examples to Level Up Your Prompt Engineering
Not only ChatGPT, but you can apply the following prompts example/context to any AI tool that’s, in particular, a language model, for example, Bard, Bing AI and Jasper.Ai to improve your prompt engineering.
If you’re a beginner using the ChatGPT or other AI tool, you can prompt the tool like this,
“Hey, ChatGPT, I am new to the ChatGPT interface. Could you please guide me through it first, before we do an actual conversation?”
Or, if you’re going after a complex topic and don’t want to confuse yourself with intricate terms, you can prompt ChatGPT like the following,
“Explain [your topic name] in simple terms. Explain to me as if I’m a beginner.”
Or, if you need information for someone else, you can prompt ChatGPT like the following:
“I need information on [your topic], but it has to be simple, neither beginner nor advanced.”
You can also leverage the 80/20 rule by mentioning it to the AI tool while gathering the information. This is how you can do it,
“I want to learn about [your topic]. Ensure you provide me with the 20% information that would help me learn the 80% of the topic. Follow the 80/20 approach.”
And if you want to learn a new skill and approach it in a beginner way, you can go like this: “I want to learn HTML. I am a complete beginner. I want to do it in 30 days, so create a plan that will help a beginner like me to learn and improve the skill.”
You can also provide the topic and ask the AI tool to generate content ideas. For example, you can prompt, “Topic: How to Boost Content Marketing Through AI tools? Generate some unique and innovative content ideas that are not common for the above topic.”
The above prompts can help a beginner make the most of AI tools.
With all this information on how to write a prompt, you can guide your AI tool to generate responses that would fulfill your requirements.
When you utilize the points mentioned in this blog, you’ll notice a difference between your previous prompts and their responses; you’ll find how effective the response you’re getting from the language models. Once you start practicing these, you’ll reach the peak of prompt engineering. You’ll level up your skills each day and become a successful prompt engineer.
I hope these points have helped you understand how to use AI tools most effectively. If you have any thoughts or something you’ve experimented with AI tools, then don’t forget to mention it in the comments.