1
“Give me a second, I’ll do it with ChatGPT.” Be honest, how many times have you heard this phrase in the past two years? Maybe too many to count. But is it really that simple to get responses that enhance and elevate the quality of our work? It’s not as easy as asking a question and magically receiving a perfect answer. This illusion of simplicity often hides a surprising level of complexity.
Since ChatGPT entered our lives, the role of the Prompt Designer has begun to emerge across a wide range of industries.
A prompt is the request for information that we give to an AI program — a string of text that forms a question. The more generic the prompt, the more vague and imprecise the answer will be.
This is where the Prompt Designer comes in. Armed with the right technical knowledge to interact effectively with AI and a deep understanding of human language, they help translate AI responses into results that are easy for people to grasp. This is a multidisciplinary role, blending both technical and humanistic skills.
But does everyone need to become a Prompt Designer to get quality responses from ChatGPT and other AI tools? Thankfully, no.
AI programs are not comparable to human intelligence. They’re designed to calculate the probability that something is correct, but they don’t have the ability to form critical, creative, imaginative, or emotional thoughts—and that’s a good thing.
AI models, like language-based ones, are incredibly powerful at recognizing patterns, processing vast amounts of data, and making probabilistic predictions. But these abilities don’t stem from conscious understanding or self-awareness. In short, AI is “intelligent,” but only in a specific and limited sense. It excels in tasks where data processing is key, but it lacks intuition.
When we communicate with machines, we need to create a shared code—something that looks human but includes clear instructions to generate a satisfactory and accurate response. And this is where our guidelines come into play.
Imagine you’re at a casual gathering with friends in the outdoor garden of a brewery. New people join the group — typical “friends of friends.” After some introductions and conversation, you ask a guy with an exotic accent, “So, where were you born?”. Now imagine he answers, “In a hospital.” How would you react to that answer? Sure, it’s not wrong or inaccurate — he was probably born in a hospital. But something feels off. There’s an unsettling lack of human nuance. Maybe if your question had been phrased differently, as “Your accent sounds different from the locals. May I ask which country you’re from?”, the answer wouldn’t have left you feeling stuck in this awkward gap of understanding.
Creating a good prompt works the same way: the more complete and precise our question, the more the answer will foster a productive and engaging conversation.
A prompt needs to provide all the information necessary for the machine to “think” like a human. The more detail we give, the more specific we are, the more the machine will return an answer that feels close to human thought—and even empathetic.
One of the most exciting things about ChatGPT and similar tools is that we can experiment endlessly, challenging ourselves to find the perfect prompt formula.
Here’s a quick guide to start structuring a more complete prompt. Don’t worry, you can start with this and tweak it each time you copy-paste, rather than reinventing the wheel every time.
1. Give the AI a defined role or persona. Have it assume a specific role, which will help create context. After all, we wouldn’t want someone to show up at our desk and ask us to design a space shuttle (unless we happen to be aerospace engineers).
2. Clearly express the action your “character” should perform: “write,” “imagine,” “create,” “explain,” “list.” Make sure there’s no doubt about the type of output you’re expecting. For example, if you ask, “Can you write a 10-line text about [argomento]?” versus “Write a 10-line text about [argomento]”, you risk the response being simply “Yes,” leaving you in the same awkward position as in the earlier example.
3. Add details that complete your request: how many words should be used? In what format? In what language? What tone or writing style should be applied? Who is the target audience? It helps to enclose your input within quotation marks to clearly define these parameters
4. Provide examples. If you have a best practice in mind or something you’ve seen that you really like, include it as an example. Similarly, if there’s something you don’t want, be upfront about it. The AI won’t be offended.
You can be very specific, but be careful not to overcomplicate your prompt, as this could water down its clarity or introduce unnecessary complexity. Experiment with different structures that best suit the responses you’re looking for, and save your most effective prompts.
Finally, don’t forget to give feedback. Encourage the answers that satisfy you and explain why the others didn’t hit the mark. This helps the AI improve its future responses.
1. Through our experiences, we’ve noticed that AI generates more complete and creative results when asked to interpret a task rather than just execute it.
2. Google will increasingly penalize content that’s 100% AI-generated. Use this powerful tool wisely: don’t copy and paste content directly from the AI, but instead use it as a canvas to craft personalized and unique messages.
3. Don’t settle for just one answer. Ask for alternatives, or request different perspectives, especially if you’re using ChatGPT for brainstorming. You might just strike gold!
