When the Answer Is Wrong

Lesson 9 · taught by Pip

Here's a thing nobody tells you up front, friend: an AI assistant can be confident and wrong at the very same time.

It does not "lie" on purpose. It's a prediction tool — it guesses the next likely words, the way you might finish a friend's sentence. Most of the time that guess is helpful. Sometimes it sounds just as smooth and certain while being flatly incorrect. Smoothness is not the same as truth.

So your job is not to trust it blindly, and not to distrust it either. Your job is to check, then steer.

How to smell a wrong answer

Think of yourself like someone tasting soup before serving it. You do not need to be a chef. You just need to notice when something is off.

Watch for these:

• Specifics you can't verify. Exact dates, prices, quotes, page numbers, or "studies show." If it sounds precise but came from nowhere, treat it as a maybe, not a fact.
• Names and links. AI sometimes invents real-sounding sources, book titles, or web addresses that don't exist. Always open the link yourself.
• It contradicts what you already know. Trust your own knowledge of your own life. If it describes your situation wrong, the answer built on top of it is wrong too.
• Too tidy. Real life is messy. An answer with zero caveats, where everything lines up perfectly, deserves a second look.

A made-up fact stated with total confidence is so common it has a nickname: a hallucination. Same word as seeing something that isn't there.

How to steer back

When something seems off, you do not start over. You nudge. Try these, plainly:

• "Are you sure about that? Double-check it."
• "That part is wrong — I actually [the real fact]. Redo it with that in mind."
• "How do you know that? Where would I confirm it?"
• "Give me the answer again, but only include things you're confident about, and flag anything you're unsure of."

Telling it the correct fact is the strongest move. Remember, it only knows what's in front of it. If you hand it the missing piece, the next answer is usually much better.

Think of it like giving directions to a friendly stranger who's eager but a little lost. You don't yell. You just say "no, turn left here," and off you go again.

One more habit: for anything that matters — money, health, legal, a message you can't unsend — confirm it with a real source or a real person before you act. The AI is a fast first draft, not the final word.

Your turn

Open your AI tool and ask it for something with checkable facts — say, "List three quick facts about my town," or a recommendation with specifics.

Now play inspector. Pick one claim and ask: "How do you know that, and where can I confirm it?" If the answer wobbles, say so plainly and ask it to try again, this time flagging anything it's unsure about.

Notice the difference between its first confident answer and its more careful second one. That gap is exactly the thing you now know how to catch.

You're no longer just taking answers — you're checking them. That's the difference between someone who uses AI and someone who uses it well. In our last lesson, we'll round it out with the habits that keep you safe — what to share, what to hold back, and how to lean on AI without handing over what you shouldn't. 🔦