How to Write Good AI Prompts to Enhance Productivity

ChatGPT, DALL-E, and other AI programs work like magic, instantly generating thoughtful responses, creative images, and more from just a few words. But these systems are only as good as the prompts you give them. Hence, it is important to know how to write good AI prompts. AI-generated art, for instance, showcases the creative possibilities that well-crafted prompts can unlock, inspiring artists with diverse genres and new digital creations.

A well-crafted prompt is the catalyst that transforms an AI model into a powerful tool for creativity, productivity, and more. If it’s lacking, your output will be more or less the same. Remember the lines: Garbage in, garbage out!

In this article, I’ll share practical tips for structuring your prompts and boosting outputs regardless of which AI generative tool you use. I will also dive deep into our tools from the FlyMSG AI family: FlyEngage AI (For quick and easy social engagement) and FlyPosts AI (For thought leadership LinkedIn post generation).

What Are AI Prompts And How Do They Work Exactly?

AI prompts are commands you give to an AI tool like OpenAI’s ChatGPT or Google’s Gemini to make it do something—write content, interpret a statement, brainstorm ideas, review material, and so on. To write AI art prompts effectively, it is important to blend practice and creativity, focusing on key strategies and elements to craft compelling instructions for generating art.

Basically:

• You provide a line of instructions to your preferred generative AI tool. E.g. Tell me all you know about charity team building.
• The AI tool analyzes your instructions or prompts and provides an answer that’s fitting.
• You tell it if the answer needs some refining, if it meets the cut, or if it’s totally off the grid.

Source: Gemini

Fundamentally, generative AI tools rely on Natural Language Processing to interpret these prompts.

• First, they break down your prompts into fragments with NLP and analyze the context of each fragment (intent, emotion, and focus).
• Afterward, they review existing internal databases for similar patterns of questions. This is efficient because they are trained on millions of private and public human data.
• With the use of probabilistic predictions, permutations, and accuracy checks, the tool provides an appropriate answer, which you can refute and ask for a better one.

So, it’s safe to say your AI tool is limited by how much data it was trained on and the prompts you give it. While you have little control over the former, you can at least work on the latter. These programs practically follow the pattern of your prompts to generate a response. If the prompt is subpar, then your results will be the same, no matter how powerful the tool in question is.

Major Types Of AI Prompts

Not all prompts are the same despite having the same goal—to guide an AI tool into doing something for you. If you want to pass the right command appropriately, then it’s essential you understand the different types available.

Question Prompts

Remember when you’d search for things like “Does Farting Burn Calories?” on Google. Well, maybe you’ve never really asked for stuff like that. But the point is Question prompts are as simple as they can get and are usually used to define, validate a point, or request explanations for certain queries.

• What: What causes earthquakes?
• Why: Why is the sky blue?
• Where: Where can I sell my Rolex without papers?
• When: When was the internet invented?
• Who: Who invented the telephone?
• How: How to learn Latin?

Conversational Prompts

Unlike question prompts that are one-shot “ask and get” commands, conversational prompts aim to dialogue with the AI program. In this case, you’re treating it as an equal or a brainstorming partner, each sharing ideas, ruling out negatives, highlighting positives, and coming to a single conclusion.

Here’s a crude conversational prompt, for example:

Here’s what ChatGPT followed up with:

From here, we can both leverage information gathered during the conversions and reach an effective conclusion.

Creative AI Art Prompts

Creative prompts are more specialized commands that require an AI program to tap into its imaginative algorithms and produce something less generic or out of the box. For example, you could ask it to write a new song, generate sci-fi scripts, create intuitive advertising copies, tell bedtime stories, or use AI art prompts to generate images.

• Write a short story about a young girl who discovers a hidden portal in her grandmother’s garden that leads to a magical world where time flows differently.
• Compose a poem about the changing seasons, using vivid imagery to capture the transformation from winter to spring.
• Create a catchy advertisement for a new eco-friendly water bottle that emphasizes its durability, stylish design, and environmental benefits.

Data Analysis Prompts for AI Image Generator

Data analysis prompts focus on using the computing power of a generative AI to analyze, restructure, and provide a more astute interpretation of existing data. In this case, you need to first provide the data either by typing it, uploading a document, or telling your AI tool to use real-time data from the Internet.

Examples include:

• Examine the sales data above to determine the most popular menu items at local restaurants.
• Use this data to conduct a sentiment analysis of customer reviews to understand satisfaction levels with current delivery services.
• Leverage your ability to surf the internet and analyze B2C customer purchase patterns over the past year to identify seasonal trends.

In addition to the prompts already discussed, you can also provide instructional-type prompts such as translating one language to another, comparing two products, and completing a sentence.

2 Practical Steps To Write Good AI Prompts [+ Examples]

Writing good prompts is not rocket science. Once you get a hold of the guiding principles involved, it becomes easier to communicate with these AI programs and drive unique results. 

Let’s quickly explore some basic steps you should implement when writing a good AI prompt.

1. Be Clear And Specific

One big issue with AI-human communication is that we tend to believe an AI program understands everything we say, regardless of how we say it. Well, in some cases, that might be true. 

For instance, I can type this “Waht si the correct nmae of America’s Prsedent?” in ChatGPT and it will still give me a correct answer because it understood my intent despite the errors.

But that won’t always happen, especially if you’re working with complex prompts—involving more analytical power. Like humans, AI can mistake your intent for another. And that means you’ll end up getting the wrong results.

So, the first step to writing good prompts is assuming AI programs are your next-door human neighbors who are not omnipotent. They are not in your mind, and they can make mistakes. 

In other words, you’ll need to be more straightforward and as layman as possible with your prompts. Here’s an example of similar prompts but differences in specificity:

Prompt 1: Write a 50-word summary of the plot of “To Kill a Mockingbird”

Prompt 2: Write a summary of a book.

Both prompts generated responses, but the first prompt was clearer, mentioned a specific book, and indicated how many words were needed. This guideline guides the AI program towards providing an answer that fits, compared to the second prompt, which was totally ambiguous and directionless.

2. Provide Context If Necessary

Remember we said you have to treat AI as humans—they are not perfect. That’s why you need to provide more intricate details when writing your prompts, not just dumping three-word prompts and requesting non-generic answers. That’s not going to work.

Context makes your prompt clearer, portrays your intent, and helps the AI program understand exactly what you need. Anything contrary to this will only lead to responses that are out of your expectations or requirements.

A perfect example is from OpenAI’s Sora. 

Sora’s Prompt: A stylish woman walks down a Tokyo street filled with warm glowing neon lights and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.

Here’s the result.

Source: Sora

The same goes for digital art generators like Dall-E and text-generating AI programs like Gemini and Claude or more content-specialized ones like Jasper. Provide as many details about the subject as possible when you’re generating AI art or before hitting your “generate” button for textual content.

Check out these art prompt examples for your preferred AI image generator:

• Futuristic Cityscape with Neon Lights: Imagine a bustling metropolis in the near future, where skyscrapers adorned with neon lights stretch into the evening sky. Futuristic vehicles hover or glide along illuminated streets, casting colorful reflections on sleek, polished surfaces. The atmosphere is a blend of modern technology and urban vibrancy, with a hint of cyberpunk aesthetic.
• Enchanted Forest with Mystical Creatures: Picture a secluded, ancient forest untouched by time, where sunlight filters through dense foliage to illuminate a world of magic. Ethereal lighting bathes mythical creatures like unicorns, fairies, and perhaps even dragons in a soft, enchanting glow. Moss-covered trees twist into whimsical shapes, while sparkling streams and mysterious glens beckon adventurers into a realm of fantasy and wonder.

AI art generators work by using deep learning algorithms trained on vast datasets of images to interpret these prompts and pop your art up.

If you’re writing an article, you can follow formats like this, specially when using AI writing assistants:

• Topic and Scope: Write an in-depth 800-word article on the impact of artificial intelligence on customer service experiences in e-commerce. Focus on how AI-powered chatbots and personalized recommendation systems enhance user satisfaction and conversion rates.
• Key Points to Cover: Discuss the benefits of AI in handling customer queries efficiently, improving response times, and personalizing shopping experiences. Explore case studies from leading e-commerce platforms to illustrate successful AI implementations.
• Audience and Tone: Target tech-savvy entrepreneurs and marketers looking to optimize customer engagement strategies. Maintain a professional yet accessible tone, balancing technical insights with practical implications for business growth.
• Sources and Evidence: Include statistics and quotes from industry reports or experts to validate the advantages of AI in boosting sales and customer loyalty. Ensure the content is current and supported by credible sources.
• Conclusion and Call to Action: Summarize AI’s transformative role in shaping future customer service trends. Encourage readers to consider integrating AI technologies to stay competitive.

You can add as many more as possible and proactively shape how you’d want the output to be.

How can I adapt AI prompts for different AI capabilities?

Different Artificial Intelligence systems vary in capability, therefore requiring customization. Take into account the key features of AI and tailor the response which uses them most efficiently. Do your research in advance to see if it is suitable for your application.

Consider perspective, lighting, and composition

Further detail – including lighting source, frame, resolution – may affect your picture’ s quality. If your picture has a very busy background, then add a blur to it. You should also try re-formatting your picture for more dimensional impact. Some examples of how to put those refinements in words include: a watercolor painting by a samurai warrior in the style of Caspar David Friedrich with rough canvas textures.

Define the composition

Finally, there are some keywords that define your images composition. These include things such as resolution, lighting styles, aspect ratios, and camera views. For example, this midjourney results show: This example shows how images can vary according to the prompts.

Add more details to your prompt

Are you into making things even better? The topic and the art form can be added with additional input in your prompts which influence the final result.

Advanced tools and strategies for AI art prompts

Here’s some advice for you to achieve legendary “imaginator” status: Reverse engineer the actual images to discover fresh questions and answers. CLIPI-interrogator is an AI Model hosted by Hugging Face. It takes an image and prompts you according to its content. The results are useful in building detailed prompts for converting text-to-image models. Across and in Pictures. Text in images is only a possibility. You can even use images to create inspiration and you can also tune it through Text / Model Control.

Let It Know The Tone You Want

“Tone” is how you want to convey a piece of information or how you want people to see it. For writers, this includes the style of writing used, the emotions involved, the choice of words, and the demeanor with which each sentence is crafted. That’s why it’s rare to see two writers writing the same way.

For businesses, tone describes a brand in its entirety. So, if your brand is in the children’s beauty and fashion industry, you’d most likely choose anything “playful” to engage your audience. 

Besides being playful, tones can also be formal, informal, conversational, optimistic, authoritative, playful, empathetic, persuasive, academic, sarcastic, and so on. Once you know the one that best aligns with your interest, include it in your prompt.

Example Prompt: Write a persuasive 50-word article advocating for renewable energy adoption, using an optimistic and hopeful tone to inspire readers about the potential for a sustainable future.

Otherwise, you can follow step 3, which emphasizes providing context. In this case, you can write a short paragraph and ask the AI program to follow your tone in the provided paragraph when generating its own texts.

The more context you provide, the better.

Use “Act As If” Commands

Every AI sees itself as an AI. That in itself limits creativity and can make them generate value-less responses. You need to tweak that view and assign an identity to whichever AI program you’re using.

For instance, you can ask ChatGPT to assume the role of a human copywriter. This means embodying whatever qualities a human writer has—passion, brand empathy, mastery of language, authoritativeness, and sense of self.

Of course, AIs can’t perfectly mimic us at the moment, but they can still fake these instances to produce better results.

Prompt: Act as if you’re a human copywriter and give me a 20-word long copy for my Twitter ad. The product is a pink face scrub, provides smooth skin, and targets age 14 to 22 female groups in California. Speak to them directly and appeal to their needs.

Depending on your needs, you can request the assumption of different identities in your prompts and also provide a context to replicate in case you need more specific results.

Include Helpful Examples

Context provides more details, and Act Ifs provide identity, but that might not necessarily be enough to generate down-to-earth responses from your AI. This is especially true if you’re prompting it to do something much more complex, like creating case studies, generating an image, or putting together a short clip.

In such situations, providing examples of what you want is much more helpful. 

For instance, you can upload a case study PDF to ChatGPT just to help it understand the type of result you need when you request a similar output. Or copy-paste an entire article in case you embrace AI for blog post generation.

Reference Previous Prompts

Quite a good number of people turn to AI to accomplish even the simplest tasks every day. That means we get to experiment with tons of prompts each time, and some are unexpectedly better than others.

There’s no hard-and-fast rule against visiting those prompts that yielded better results and reusing them again. Copy, insert your exact needs, and paste for a result. Of course, referencing or reusing a previous prompt that once performed well is not a foolproof method.

So, you still have to apply the other steps we’ve discussed above to tweak your final prompt.

Encourage Creativity With Open Prompts

While boxing your prompts lets AI programs provide a more specific result, it also limits creativity if you overdo it. In cases where you need ideas, different structures, and something outside the box, you should instead opt for open prompts.

This involves using broad and flexible instructions that allow for a variety of imaginative responses. No constraint on formats or outcomes, thus leaving space to explore different ideas or approaches.

A crude method: You can also ask your AI to generate prompts that are likely to work best on its system.

Prompt: Give me 3 suitable prompts for AI art or provide useful art prompt ideas for my designs. I want to use it on Dall-X, a free AI art generator, so ensure it is tailored to the platform.

Consider Variability And Give Feedback For Improvement

Reiterating this again: AIs are not omnipotent. Providing context and all can help them grasp a good percentage of your original intent, but not 100%. So there’s bound to be some variations occasionally. That’s, however, not a problem since you can easily edit their output to suit your needs.

Going forward, you need to consider the different variables that can affect the results you get and adapt to them in the next prompt. Some AI programs like ChatGPT and Gemini also encourage instant feedback, which means you can let their algorithms know if they got it right or not or how they should have responded better.

Wrapping Up

AI programs are magical innovations, except that they are not run by magic. The results you get are as good as the prompts you give. If the prompt is poor, the outcome will likely be as poor. And if it is great? Well, you know already.

To supercharge your AI prompts, embrace clarity and be specific by providing detailed context. You can also provide examples where necessary and use Act Ifs for identity. Other methods include referencing previous prompts, encouraging creativity with open prompts, and considering variables no matter how good your prompts are.