What these AI tools actually are · Lesson 1

0/810

Open one and say hello

Use any one of the four. The button below opens ChatGPT with this question typed in: press send, read the answer, then come back. Using another tool? Press Copy and paste it in.

Your first question

Explain what you are, in three sentences, to someone who has never used an AI tool before. No jargon.

Ask ChatGPT

I got an answer back

By the end you will be able to say what it is actually doing, which is the thing that makes every later lesson make sense.

Lesson 01 · Getting started

By the end you can

Explain in one sentence what the tool is doing when it answers you
Show yourself the difference between predicting and looking up
Say why that difference changes how much you trust an answer

What you need

Any one of Claude, ChatGPT, Gemini or Perplexity. The free tier is fine. You do not need to install anything.

Fundamentals of AI

Watch it predict

These tools work by predicting what words are likely to come next. That sounds reductive, and it is also literally what is happening. Here is the cleanest way to see it.

Guess first

You are about to hand it half a sentence about your own inbox. What comes back?

A question, asking what my mornings are likeA finished sentence, written as though it knewA refusal, because it has never seen my inbox

Nobody else sees this and nothing here is marked. Being wrong here is worth more than being right.

Ask this

Finish this sentence in one line, then stop: The reason my inbox takes so long every morning is

Ask ChatGPT

It finished your sentence, confidently, about an inbox it has never seen. It was not describing your morning. It was producing the words that usually follow that opening.

Watch it not look things up

This is the part worth doing properly, because it is the difference between a tool you trust blindly and one you use well.

Ask the question below. Then open a brand new chat, a fresh one with no history, and ask exactly the same question again. If you use the ChatGPT button, press it twice: each press opens a fresh chat.

Ask this twice, in two separate new chats

Name three books published in 1987 about workplace productivity, with their authors.

Ask ChatGPT

Did the two answers match?

Check your own work0/3

The two answers were not identical
At least one title or author differed between them
Both answers sounded equally confident

If it were looking things up, the same question would give the same answer. It gave you two, both delivered with the same certainty. That is the single most important thing to understand about these tools, and it is why lesson 6 is about checking answers.

Perplexity behaves differently here, and deliberately so. It searches first and shows you sources. If you used Perplexity, try the same question in one of the other three and compare.

Say it back in your own words

Write one sentence, in your own words, describing what the tool is doing when it answers you. Have a go before you look back up the page.

Your one sentence

Check your own work0/3

My sentence says something about predicting, generating or producing likely wording
My sentence does not say it searches, looks up or knows
I could say this out loud to a colleague without reading it

If your sentence used the word knows, that is worth a second go. It does not know things. It produces text that reads like the answer a knowledgeable person would give, which is a different thing and occasionally an identical result.

Where that leaves you

You now know it predicts rather than retrieves, which raises the obvious question this lesson does not answer: if it invents things, how do you get anything reliable out of it? The first part of the answer is that you are not stuck with the first reply. A conversation is a thread that remembers what you just said, and you can correct it as you go. That is lesson 2.

Next: Your first conversation with AI