It can be wrong

This is the most important lesson in this whole course. Read it twice.

Hallucinations

AI can produce things that *sound* completely correct and are completely made up. The technical name is hallucination. It happens because the AI is, remember, a guesser — and sometimes it guesses wrong with great confidence.

When it's most likely to happen

• Specific facts, dates, statistics
• Names of people, books, court cases, scientific papers
• Quotes — it will invent quotes that were never said
• Anything very recent (some models have a knowledge cutoff)
• Math with many steps

How to protect yourself

• For anything that matters — medical, legal, financial, factual — verify with a real source. Treat AI as your fast first draft, not your final answer.
• Ask: *"How confident are you in this? What might be wrong?"* It will often own up.
• Ask: *"Cite your sources."* If it can't, the information is more likely invented.
• Use Perplexity or ChatGPT's search tool when you need cited, real-world facts.

Trust, but verify. Especially when the stakes are real.