By all appearances, chatbots (ChatGPT, Claude, Llama, Grok, etc.) are just another set of black boxes of technology, glowing windows into the digital world. But unlike the fabled genies of old, they grant responses instead of wishes.
Therein lies both their charm and their challenge.
At the heart of these digital oracles is something deceptively simple: language. Not programming code, not numbers, but the everyday language we use to write emails, ask questions, tell stories, and even argue with loved ones. That, remarkably, is the stuff of thought for these machines.
How does one talk to a thinking machine?
The Language Model as a Mind
Before anything else, think about what ChatGPT, for example, actually is. It is a Large Language Model (LLM), a system trained not to know, not in the way humans know, but to predict what comes next in a sentence. If I say, “The sky is…”, ChatGPT, like an attentive student of language, will offer: “blue,” “overcast,” “beautiful,” or even “falling,” depending on the context you’ve given.
It has no beliefs, no memories of the past unless you provide them, and no understanding of truth in the way we commonly hold it. But if you speak to it clearly and thoughtfully, it can generate responses that are astonishingly helpful.
Discovering the Dialects of Machines
One challenge is to recognize that not all chatbots, or even various models or versions offered by various chatbots, are alike. Some are faster, others more detailed, and some—like the newest models—are capable of juggling text, images, even voice. Choosing a model is like choosing a partner for a dance: know their strengths, anticipate their limitations.
We call this “discovering the models,” and it’s the beginning of the art of collaboration with AI. A newer model may better understand nuance; an older one may be quicker and more cost-effective. Learning to pick the right model is the first step toward fluency.
Prompting: The Skill of Saying What You Mean
Let’s assume you’re using ChatGPT. Now that you’ve chosen your dance partner, how do you lead? You start with a prompt.
Prompting is not merely typing in a request. It is crafting an instruction with intent. Like asking a librarian for help, you must be clear: do you want a summary or an analysis? A friendly tone or a professional one? One paragraph or three?
You might say, “Tell me about climate change.” That is a start. But if you say, “In three bullet points, explain the causes of climate change as you would to a curious 14-year-old,” you have handed the machine a compass, not just a question.
This is the craft of prompting. It is part science, part art, and entirely learnable.
Providing the Mind with a Map
Then comes the matter of context, perhaps the most underestimated element in all of human (and now machine) conversation.
Imagine calling a friend and saying, “We’re still on for it, right?” Without context, they’ll fumble. The same is true for ChatGPT. If you don’t tell it who it is, what it’s doing, and for whom it is writing, it will guess. Sometimes badly.
Setting the context is like laying down the first page of a play script: “You are a health advisor writing to new parents. Use a warm, concise tone. Provide three tips about infant sleep habits.” That first message isn’t just a prompt. It’s the overture.
Speaking in Documents and Pictures and Voice (Oh My!)
Of course, ChatGPT is not confined to text. You can feed it documents, pictures, and even your voice. You can ask it to summarize PDFs, analyze diagrams, or read your spoken questions aloud and respond in kind.
This is not just a search engine or an auto-complete assistant. It is a multi-modal assistant. Learn to attach files, upload screenshots, or speak directly, and you’ll find new dimensions of collaboration with this system.
In Closing: A Partnership of Clarity
You are not programming a machine. You are collaborating with one. Clarity is the currency of that collaboration.
This is why, in our workshops and (coming soon!) in our guided learning modules, we begin with some core modules designed to build your intuition and establish a framework on which you can build. You will learn to choose your tools, speak clearly, frame your context, and share inputs in new ways. Modern chatbots are not magical, but they are increasingly powerful. Power, used well, requires precision.
Just as a telescope reveals the stars only when you know where to point it, ChatGPT reveals its brilliance (and augments yours) only when you learn how to aim.
Much of the material in our blog is intended to help you learn how to aim and what to aim for.
Welcome to a new adventure of learning!
Leave a Reply