PART I — Learning How to Work With AI
Chapter 2
What Are We Actually Talking To?
The Name on the Screen Doesn't Tell You Everything
When you use ChatGPT, Claude, Gemini, or another AI assistant, it is natural to think of the name on the screen as the thing you are talking to.
It isn't quite that simple.
The device in your hand is how you communicate with AI. What happens behind that conversation can involve models, information sources, computing systems, and other services.
You don't need to learn all of that technology before you can use AI.
But having a useful mental model of what is happening can change how you think about the answers you receive.
AI Isn't One Single Product
You may use several AI assistants and discover that they behave differently.
That doesn't necessarily mean one is "real AI" and another isn't.
Different assistants can use different models, tools, information sources, and connected services. They may approach the same question in different ways.
The important question isn't simply which assistant is best.
It is which assistant is best suited to what you are trying to accomplish.
Fluent Doesn't Mean All-Knowing
One of the most striking things about AI is how naturally it can communicate.
It can explain an idea, answer a question, rewrite something, or carry on a conversation in language that feels remarkably human.
That can create a misleading impression.
AI can communicate fluently without knowing everything. An answer can sound convincing and still be wrong.
Learning to separate how well an answer is expressed from how trustworthy the answer is becomes an important habit.
What You'll Discover
Chapter 2 develops a mental model for thinking about what sits behind an AI conversation.
The book goes further into:
how an AI model differs from the device you use;
why different assistants can produce different results;
how AI generates responses;
why fluent language can create false confidence;
and how to think about AI without needing to become a computer scientist.
You don't need to learn the machinery to use AI well.
You do need enough of a mental model to know what kind of conversation you are having.
Try It Yourself
Have a conversation with an AI assistant about what it is.
Ask it to explain, in language you can understand, what happens between the moment you send a question and the moment an answer appears.
Then ask it what parts of its explanation could give you the wrong impression.
The point isn't to memorize technical details.
It is to begin questioning your assumptions about what is happening behind the conversation.
Why Read the Book?
Chapter 2 establishes a mental model that becomes useful throughout Understanding AI.
Once you recognize that the device, the service, the model, and the information surrounding a conversation can all be different pieces, many later questions become easier to ask.
The book builds from that foundation into prompting, conversation, research, decisions, learning, creativity, everyday work, and the larger questions raised by increasingly capable AI.
Knowing what you are talking to is a good place to start.