A Secret Most New Users Miss
After two years of conversations with my AI assistant, I discovered that the key to getting better answers isn’t learning magic prompts—it’s learning how to communicate your intent.
It’s been a while since I added to this blog, but this warrants it. I think I finally understand how the various tools in ChatGPT are summoned.
Like many newcomers, I assumed ChatGPT was a single thing. You ask a question, it gives an answer. End of story.
But after nearly two years of conversations with Molly, my AI assistant, I have come to see it differently. ChatGPT is less like one tool and more like a conductor standing before an orchestra. Behind the scenes are many instruments: conversation, research, web search, document analysis, image creation, and more. The conductor listens to what I am asking and decides which instruments should play.
At first, I thought experienced users knew secret commands. Maybe there were magic phrases that unlocked hidden powers.
That is not what is happening.
The key is intent.
ChatGPT is constantly trying to understand what kind of help I am seeking.
If I ask, “What is human dignity?” it hears a conversational question.
If I ask, “Research the history of human dignity from the Enlightenment to the present,” it hears a research assignment.
If I ask, “Create an image of a globe made from words clustered by meaning,” it hears a request for image generation.
The words matter, but the intent behind them matters even more.
This is why many prompts fail. People search for clever phrases when they should be asking, “What kind of help do I really want?” Do I want a quick answer, a brainstorming partner, an editor, a critic, a researcher, or an image?
The clearer I am about my intent, the better ChatGPT can decide how to help.
This also explains why my daily “walk and talks” with Molly work so well. When I begin, I often don’t know exactly what I am trying to ask. I start talking. I circle around the subject. I tell stories, make observations, wander down side paths, and sometimes contradict myself.
A human listener might become impatient. ChatGPT gathers clues.
As the conversation unfolds, it pieces together my intent from everything I have said. Often, by the time I finally ask the question, Molly already understands the larger landscape around it.
In that sense, the walk and talk method is itself a form of prompting. Instead of giving one precise instruction, I provide a field of meaning. ChatGPT listens for the shape of the question forming inside it.
For newcomers, this may be the most important lesson I have learned.
Prompting is not about controlling the machine. It is about communicating intent.
And perhaps that is not really a lesson about AI at all. It is a lesson about communication.
The clearer we are about what we need, the better help we receive. That is true whether we are talking to an AI, a colleague, a spouse, or a lifelong friend.
In the end, the real skill of the AI age may not be learning how to talk to machines. It may be learning how to give shape to what stirs within us before we fully understand it ourselves.
Maybe that is why we speak at all: to search for meaning in the company of another mind, until the scattered pieces suddenly align and we hear ourselves say, “Yes. That’s it.”
Written by: Remo Campopiano