When you hear Artificial Intelligence (AI), you might picture robots, chips or code. Put those images aside for a moment.
Imagine walking into a bag boutique and saying: “I dress pretty simply. I want something I can carry to work but that will not feel too formal on weekends. My budget is about two thousand dollars.”
An experienced sales assistant may not need fifty questions. She can probably show you three options that are much more likely to fit.
Why? Not because she can read your mind. She has simply seen a lot of customers before.
AI is good at finding patterns in many examples
After years on the shop floor, a good assistant notices patterns. Commuters often care about weight and capacity. People who say “not too formal” may avoid very structured business bags. Minimal dressers may prefer cleaner shapes.
At the most basic level, AI does something similar: it uses a lot of data to learn patterns, then uses those patterns to make predictions, decisions or generate results.
Real AI does not think, “She feels like a taupe bag today.” It uses numbers, computation and a model. But “learning patterns from many examples” is a very useful place to begin.
Does that mean AI thinks like a person?
No.
A human sales assistant has emotions, life experience and social understanding. She may notice your expression, remember your wedding next week or realize you secretly dislike a bag even when you are being polite.
AI can look surprisingly understanding because it is very good at using patterns in the information you give it. That does not mean it experiences the world in the same way you do.
When ChatGPT sounds thoughtful, a useful mental model is:
It is very good at using the clues in front of it to predict what kind of response is likely to make sense next.
That is more useful than imagining a tiny electronic person living inside the computer.
AI is a huge umbrella term
“AI” is a little like the word “bag.” A tote, clutch, backpack and evening bag are all bags, but they are designed for very different jobs.
AI is the same. Detecting a cat in a photo can be AI. A bank spotting suspicious card activity can be AI. Netflix predicting what you might watch next can be AI. ChatGPT generating text is AI too.
They do not all use the same method and they are not all good at the same thing.
So when someone says, “AI can do this,” ask one more question: What kind of AI, and what was it built to do?
One thing to remember today
AI is a system that can learn useful patterns from data and use those patterns to produce a result.
In Lesson 002, we will separate “AI” from the model that actually does a particular job.