How AI Models Work

The people and companies in this course are made up. Check any number against your own situation before you rely on it.

A language model generates responses from learned patterns and the context it receives. Some products also search the web or run code. Check which tools were used and inspect their results. A confident or fluent answer isn't proof it's right, and neither is a hedge like 'I think'.

What this means for you

• Treat output as a draft, not a fact.
• Ask for sources and check them.
• Give it the facts you want it to use.
• Expect different answers to the same question.

Example (fictional)

Ask a model for a source on a niche topic and it may give you a real-looking citation that doesn't exist. It predicted what a citation should look like. The fix isn't to avoid the tool. It's to check anything you plan to rely on.

Common mistakes

• Treating a confident answer as a correct one.
• Assuming the model looked something up.
• Expecting the same answer every time.

Exercise

Ask a model a question you know the answer to in a narrow area. Note what it gets right and wrong.