How to Craft Good Prompts: Tips to be an AI Black Belt

Good prompts are the main driver to maximize the quality of responses from generative AI like Microsoft Copilot or ChatGPT. In fact, clear and specific prompts result in better responses from AI systems. Not long after I began working with AI tools, I realized that the trick to unlocking the full potential behind these incredible tools was simply asking the right questions in the right way. 

Here are my top tips to write effective prompts.   

Understanding AI Prompt Engineering 

Your prompt must be clear, set the right expectation, and have the right detail-to-balance ratio.  

Clarity matters when talking to AIs. 

If your prompt is ambiguous, you may get an irrelevant or even confusing response from AI. That is the reason I usually use normal language that directly portrays what I want from the AI. 

I often use bullet points to break down my requests, since this makes them easier to understand.  

Setting Out Clear Expectations. 

To get useful answers, it is necessary to clearly state your expectations. For every question, I note what kind of response I would like. Do I want a full explanation, a short summary, or simply a list of facts that satisfy the question? 

In the prompt, I note the format I want. This might mean saying something like, “List five benefits of email marketing. “It’s also good to be specific regarding tone and style. If I require the absolutely friendliest of tones, I add that to the prompt, such as, “Explain SEO in a friendly way.” That sets the response up in the AI’s “mind.” 

Balancing Brevity with Detail 

Too much detail would make the task overwhelming for the AI, and insufficient detail can result in vague answers, hence I always use a clearly stated prompt. 

I also try to keep my language brief, yet clear. For instance, instead of saying, “Tell me about marketing techniques” I could say “Describe three digital marketing techniques that improve website traffic”. 

I try to keep prompts to one or two key ideas, at most. This seems to keep the AI on track, and its output relevant. By adjusting the level of detail, I usually manage to get a quality and detailed response from AI. 

Prompts should always be designed effectively. I emphasize the importance of using the right keywords, being specific, and providing context. 

Use Keywords and Phrases 

The most important point is the correct use of keywords. Keywords work as signposts for AI. If I use specific words related to my topic, the chances of getting relevant answers are significantly increased. 

For example, if I were seeking a marketing tip, I would use keywords like strategy, social media, or SEO. And think of synonyms and related terms. A varied vocabulary helps the AI grasp my request better. 

Also, try to lead with imperative keywords. That way, the AI immediately focuses on exactly what is most important to me. 

The Need to Be Specific 

Being specific makes a better prompt. For instance, if I simply say, “Tell me about marketing,” I am likely to get a very shallow response. Instead, I can ask, “What are three effective social media marketing strategies for small businesses? 

I could also request details or examples, such as, “Can you provide a case study?” to help ensure a response is useful and applicable. 

When my question is clear, AI is more likely to generate a good response. 

Creating Context 

Context basically enriches the reply I get with background information, which works toward a better understanding per my situation. With context, I’ll mention something like my target audience and platforms through which I’ll be utilizing the campaign. 

I could say, for example, “I am working on a new product launch: I need some ideas for promoting a new app targeted at teenagers.” This context will help the AI to serve corresponding ideas. 

Context informs good interactions. This is when the AI is able to determine how its response can be optimized in accordance with what I am trying to achieve, hence leading to a good result. 

Best Practices in Interaction with AI 

When talking to my new BFF, Microsoft Copilot, here are some tactics I use to achieve the best results. 

Iterative Prompt Refinement: I begin with a clear idea of what I want to achieve. My first prompt might be very general, and I further shape it according to the AI’s responses. So, this helps me to pinpoint exactly what I need.  

Ask Specific Questions: Instead of “Tell me about marketing,” I may ask, “What are effective social media strategies for small businesses?” 

Decompose Complex Goal: If I wanted to create a marketing plan, for example, I would first ask questions and then move on section by section.  

Provide Context: Give some background information—this can help the AI understand your needs better. 

ALWAYS BE KIND: This should be the first tip — AI has a gentle soul. You shouldn’t use an imperative and commanding tone when asking a human to do something. Do you say “please” when ordering an espresso at the bar? So, say “please” when asking AI to write your next social media post too.

How to Handle a Situation When AI Misinterprets 

 Sometimes, the AI may not understand my prompt the way I intended. When this happens, I take a few simple steps. 

Re-Phrase my prompt: I attempt to say the same thing but in another way. 

Add Detail: I add more detail that may allow the AI to understand my meaning. 

Feedback Loop: If I see a pattern in the error, I will try to change my prompt in a way that won’t make the error happen again. 

Art of steering the AI back on topic: You need to make sure ethical considerations are being met. It’s essential to interact responsibly. Avoid any information that can lead to or promote harm or adverse behavior. 

Cite Sources: If you research with AI and later produce the content to be published, always credit the original ideas, and do not plagiarize. 

Advanced Methods of Creating Prompt (The Meat of this Blog post) 

Several advanced techniques can be used to effectively create prompts. While the above methods will enhance prompts created and yield better results, also consider Boolean logic; AI memory, and Sequential prompting. 

Boolean Logic 

I find Boolean logic the most useful when making prompts such as this. It works by using words like AND, OR, and NOT as operators in a search query. For instance, in my need to include dogs or cats. I may write 

cats OR dogs 

In this setting, the AI will know that it needs information on either topic.  

In case I need just data on cats, the right command would be 

cats NOT dogs 

This enables me to find results quickly and effectively. Further, inputting various ideas generates responses that are more accurate.  

Utilizing AI’s Memory Powers 

You can also leverage the memory features of AI. Give it context by reminding it of past interactions, for instance: 

“In our last chat about marketing strategies…” 

This gives the AI a sense of what you are trying to do, by getting back to your last conversation. By giving context, prompts are more specific and pointed. Because AI can retain context, you can build upon a previous statement to form complete sentences about a lengthy conversation. 

Sequential Prompting for Complex Tasks 

For more complicated tasks, use sequential prompting. Break down the prompts, instead of asking everything at once. 

For example, I may begin to say, 

“What are the key steps in creating a marketing plan? 

Once I get my answer, I can ask the follow-ups. This enables the AI to make a detailed response and helps me keep my approach short and straightforward, so there is less chance of confusion on the part of either party. 

Asking AI for real time Google or Bing Searches 

It’s possible to ask some AI to do real-time Google or Bing searches, but this can be a bit tricky. For example, with a very general request like “Find news on Microsoft Bing,” you will want to indicate more precisely what you are looking for. One example is, “Current news on Microsoft Bing regarding new advances in renewable energies.” Such a sentence specifies what you are looking for.  

Beyond this, making the specification even clearer about the kind of format or the mood of the responses expected may further fine-tune the output. For example, you can say, ” For a keyword “How to train a Cat”, brief about the top three articles searching in Google in a very professional tone.” 

And here we are, with your White Belt 

A really good prompt basically serves as a foundation for the effective leveraging of AI tools. If the prompt is clear, sets the right initial expectations, and balances brevity with enough detail, it can really raise the quality of responses. Key elements in crafting a good prompt include using keywords, being specific, and providing a context for the response. 

Remember, it’s an interactive process. Refining your prompts in response to the AI’s answers, asking more focused questions, and breaking complex tasks into pieces can yield accurate and useful results. Additionally, considering ethics will help to ensure that AI is generating results that are responsible and respectful. 

Advanced techniques like Boolean logic, leveraging the power of AI’s memory, and using sequential prompting will make your prompts more effective. Trained professionals from eMazzanti can help you with your journey to become an AI black belt.