What AI Really Is (and Isn't)

If you've ever felt that everyone else "gets" AI and you somehow missed the memo, you're in good company. The good news: you don't need to understand the math to use AI well. You need one accurate mental model, and by the end of this lesson you'll have it.

The short version

The AI assistants you've heard of (Claude, ChatGPT, Gemini, Microsoft Copilot and others) are built on something called a large language model, or LLM. An LLM has been trained on an enormous amount of text: books, articles, websites, code and conversations. From all that reading it learned the patterns of language so well that it can write, explain, summarize, translate and reason through problems in plain English.

Then it was trained further to be a helpful assistant: to follow instructions, answer questions and hold a conversation.

That's it. No magic, no consciousness, no little person inside. But don't let "it's just patterns" fool you either. Those patterns capture a remarkable amount of human knowledge and skill.

A mental model that actually helps

Think of your AI assistant as a brilliant, extremely well-read new assistant on their first day:

• They've read more than any human ever could, across almost every subject.
• They work in seconds, never get tired, and never mind redoing something ten times.
• They've never met you, so they don't know your job, family, preferences or goals unless you tell them.
• They sometimes state things confidently that are simply wrong, and they won't always warn you.

Every good habit in this course flows from that picture. Because they've never met you, you'll learn to give context. Because they're fast and tireless, you'll learn to ask for changes instead of settling for the first answer. Because they can be confidently wrong, you'll learn when and how to check.

What AI is great at

Notice the pattern: these are all language tasks. If a job involves reading, writing, explaining or organizing words, AI can probably help.

What AI is not

• Not a search engine. Unless your assistant searches the web for you (many can), it answers from what it learned in training, which stops at a certain date.
• Not a database of facts. It doesn't look answers up in a verified source. It generates the most plausible response, which is usually right but not always.
• Not a mind reader. Vague requests get generic answers. What you get depends heavily on what you give.
• Not a professional. It can help you understand a medical, legal or financial topic, but it can't examine you, see your full situation or take responsibility for advice.
• Not your memory. Most assistants start each new conversation fresh, unless you've turned on a memory feature.

A quick example

Here's a typical first attempt:

Give me dinner ideas.

You'll get a reasonable but generic list. Now compare:

I'm cooking for two adults and a 6-year-old who hates anything spicy.
We have about 30 minutes on weeknights and we're watching our budget.
Suggest 5 dinner ideas. No pasta, we already eat it a lot.
For each one, give a one-line description and the main ingredients.

Same assistant, a far better answer, because you treated it like a capable new assistant who needs a briefing. You'll master that skill in Module 2.

Try it now

1. Open any AI assistant you can access. (Don't have one yet? Lesson 3 walks you through setup. Come back to this exercise afterwards.)
2. Paste this prompt and read the answer:

I'm completely new to AI assistants. In plain English, explain what you
are, what you're good at, and what you're not good at. Then ask me three
questions about my daily life so you can suggest how you might help me.

3. Answer its three questions honestly. Notice how its suggestions improve once it knows a little about you.

Common mistakes

• Treating it like a search box. Typing three keywords gets you three keywords' worth of help. Write the way you'd talk to a person.
• Trusting it blindly, or dismissing it entirely. Both are mistakes. It's a powerful tool that needs a human in charge.
• Giving up after one answer. The first reply is a starting point, not the finished product.

Key takeaways

• AI assistants are large language models, trained on vast amounts of text to understand and generate language.
• Picture a brilliant, tireless new assistant who has never met you and is sometimes confidently wrong.
• They're strongest at language tasks: drafting, rewriting, explaining, organizing, brainstorming and summarizing.
• Better context in means better answers out, and important facts always deserve a check.