When it comes to AI-powered content generation, the magic starts with a simple yet crucial element: well structured and thought-out AI prompts. But if you look at what most people use as prompts, it tends to be a few lines of simple instructions, and the expectation is that they will get outstanding results. More often than not, what's generated is of little use.

In this digital age, AI prompts are the guiding stars for Artificial Intelligence tools, helping small businesses, solopreneurs, independent workers, and content creators craft compelling content faster and at less cost. Unfortunately, most people do not know how to communicate with or guide these tools.

Further down we will dig into what great prompts are composed of, so you can start using the framework to begin whipping up your own highly effective instructions.

But firstly let's look at what AI prompts are, and why should you care about them.

What are AI Prompts? ⚙️

You can visualize AI prompts as being the steering wheel of an AI-driven content car, guiding and piloting to reach a specific destination or outcome.

Prompt engineering which we hear about so much these days, is basically the way we communicate with AI, and the prompt is the input and instructions with the goal of a desired output or result. So, we as “AI Tool handlers” need to get better at communicating with our AI assistants/co-authors.

Up until now, we have predominantly been using text prompting to communicate with our AI tools, although recently Open AI has launched a new version of their service where you can speak and ChatGPT will convert the spoken word into text! However, this development is still very new, so we will write about this soon, but for the purposes of this article, we will stick to typing text prompts.

💡

Prompt engineering is the skillset of talking to bots

Why Are AI Prompts Important? 🦾

AI prompts hold the key to unlocking the full potential of AI writing tools. They determine the direction, structure, tone, and style of your content, making them indispensable for controlling what you want generated by the AI tool. Whether you're a blogger, marketer, or small business owner, using the right prompts can save you time and effort while ensuring the content generated meets your specific needs.

How Do You Construct Effective AI Prompts? 💻

Creating effective AI prompts is an art in itself. It's like composing a symphony where each note matters.

There are 3 major high-level building blocks in crafting a great prompt:

Language – this relates to the words we use in prompting. To get great output we need great input! Remember the old saying in computer science “Garbage in, Garbage out (GIGO). The point is that we need to seriously study and string together effective commands to communicate with AI.
Models - AI looks at things in terms of models, concepts, and structures. So, when prompting it is best to provide a structure or model for the AI tool so it can return a relevant output.

Let me give you a simple example of what I mean. When I ask my AI tool to help me with an article, I may prompt it by saying:

Use the keyword “AI writing tools”, and generate a 1300-word article. The title of the article should include the keyword “AI writing tools”. Write this title in such a way that it grabs the reader's attention and generates interest in them to want to read the article further. Include 5 subheadings. Under each sub-heading, each paragraph should be composed of 200 words. One of the subheadings should be titled “ 5 best AI tools for writing”. Under this sub-heading of “ 5 best AI tools for writing”, name 5 of the most well-known AI writing tools available on the market. For each of these 5 tools, list the Pros and Cons. Also, provide a short summary of what each AI tool can do. Lastly, include an overall article summary of 150 words at the end of the blog.

The above model provides a glimpse of what kind of simple boundary to set an AI tool when prompting so that the AI software will work within these parameters. The above is still a simple prompt/model, usually I go even deeper and wider with my instructions.

Open-ended AI tools such as Chat GPT, Bard, and Claude, have access to huge amounts of data, almost equivalent to an infinite brain, so in theory, they can output many variations of a given set of prompts. To avoid receiving an output not fit for purpose, we need to be very detailed and specific.

That is why we need to provide as much guidance as possible, and these instructions or models help the AI to think…..or a better term may be to compute (because they cannot think, well, not yet!! 🫣 ).

3. Context – The 3^rd component in proper prompting is context. You need to give the AI tool some background, history, and additional information info.

Let's take cooking as an example. The model/sequence is equivalent to the recipe i.e. how much and when the various ingredients are added.

The context is how things should be done, based on previous experience. Examples of this include:

Knowing that mixing different ingredients will give a particular taste
Adding too much salt can ruin a dish
Not using a high temperature over a long period, as it could lead to over cooking

These AI systems will need the background and context to execute the different prompts to be able to produce acceptable outcomes.

So, in summary, we need to feed the above 3 main elements to help get the best output possible.

What Makes a Good Prompt? 👍

We are going to get a little deeper into prompt engineering. Below are a number of parameters we can use to generate a prompt to generate a desired output. The more of these elements you incorporate the more likely the prompt will give you the result you are looking for.

Specificity/Clarity: be completely clear, and specific plus give non-ambiguous words and instructions
Simplicity: don’t use unnecessary words
Sequential order: give the instructions in a step-by-step order
Consistency: over time provide consistent instructions, so the AI can understand the previous requests link them together, and build on what you have already told it
Action orientated: be clear on the action required. For example, create a table, with 4 columns and 8 rows.
Contextual info: give it background, some history
Assumptions - do not assume AI knows what you know
Prior existing knowledge: Refers to previous actions you may have set for it.

Prior Existing Knowledge Example: You generated 6 cold outreach emails earlier, these were sent to my database of small businesses. These emails introduced the concept of AI models and how they can help businesses. I now need a further 6 emails, they need to refer to the contents of the earlier 6 and these new emails should highlight the business benefits of using AI in business.

What Should Prompt Instructions Include? 👩🏼‍💻

Ideally, when working with an AI platform you want to set as many of the the following instructions:

Set The Character

We can set the expertise of the AI character.

Example :“You are an expert blogger with 25 years’ experience in writing articles on AI Software. You are familiar with all of the available AI writing tools and can easily help readers understand which tools would be best suited for their business needs”

Once you set this character – your AI tool will start to compute in this way.

Set The Tone of Voice

You can customize the tone using various adjectives or pick a public figure

Example: “You have the voice of an inspiring digital marketer like Frank Kern”

Set The Audience:

You can provide specific demographics of your target audience.

“The target audience is English speakers between the ages of 20 and 50 who are solopreneurs, content creators, or run small businesses”

Level of Communication:

Example: “Write the article so that a 15 year can understand the concepts”

Set The Context of Use :

Example: “This article will be published on my site https://aitoolstribe.com and will be disseminated on social media via LinkedIn, Facebook and Instagram”

Set Formatting:

Example: “Use bold text to emphasize important points and italicize quotes. Use emojis on all subheadings”

Setting The Level of Detail:

Set the level of required detail of the output

Example: “Be clear and concise with a good level of detail. Include high-level concepts and some highly detailed areas”

Rapport:

You can decide how you want to connect and generate familiarity with your audience. If your target audience is entrepreneurs, you can use words that you feel may resonate with this target demographic. I personally use words like vision, goals, metrics as I know these words are often used by my target audience.

Example: “Use visual and kinaesthetic words within the article”.

Set Phrases/Metaphors:

You may want to include specific phrases

Example: “Use the following phrases where appropriate, gamer changer”, “blue ocean strategy”, or “1000-foot view of the problem”

Set The Background Story:

Example: “I started my blog last year and this article will in a long line of articles written about AI writing tools which will be published on my website https://aitoolstribe.com. These articles will be the go-to for small business owners who want to understand the world of AI writing tools. There will also be reviews of the AI writing tools, so they can select the right tool for their business use”.

Scanning Sites For Context:

We can get the AI bot to scan other related sites to get further context and data.

Example: “ When writing this article, refer to the following websites for AI-related content: https://forbes.com, https://www.economist.com/ "

Training Models:

Setting up the knowledge base or training model can make a massive difference in the output. By training the AI generator on your specific knowledge database means that it can draw on the data that you have provided and rely less on externally available data.

💡

With an output, you can go back and refine the prompt and ask it to fine-tune the last output. You can iterate off answers, fine-tune, and get better results each time.

Prompt Engineering Summary 📖

I suggest you do a test where you write up a simple prompt for an AI tool like Chatgpt, and generate an outcome.

Then write a new but much more detailed prompt using as many of the above prompt elements as possible.

Compare both of the generated outcomes. You will undoubtedly see that the detailed prompt will give you a better result. It may not give you exactly what you want, and this is where you can use different phrasing and refine the prompt each time until you get what you are looking for. This is a skill that should be under-estimated.

High-quality prompts will get the results you want, even if you have to go through a few iterations.

If you look around the internet, many people are selling prompts, and they may be great AI prompt examples, but I do think, learning the simple skill of prompt engineering will tremendously help you in your business.

Unlocking AI's Creative Potential 🔮

Try to use as many of the prompt instructions above. How you craft and the quality of your AI prompts will directly affect the output of your AI-generated content.

These AI prompts can empower small business owners, solopreneurs, independent workers, and content creators to harness the creative potential of AI tools. Crafting effective prompts is an art that requires precision and clarity, ensuring that the generated content aligns with your goals.

Start experimenting and writing prompts, and watch your content creation soar to new heights. 🚀