AnyoneLearnAI
Start learning
Page 6 of 8~116 min topic

Why AI makes mistakes

Verify hard when the stakes are high

Not every AI mistake deserves the same energy. Match your check to the risk.

~15 min this pageDecision framework

1Try it yourself

Simulation game

Hallucination hunt

Stamp each claim: Trap or Trust. Confident voice ≠ true.

Quiz show

SOUNDS SURE

“Sydney is the capital of Australia.”

Case 1/4 · score 0

2Learn the idea

Read

A simple risk ladder

See it

Why fluent answers can still be wrong
01Predict ≠ lookupSounds like an answer
02Web is messyFacts + fanfic mix
03No embarrassmentCan sound sure
04Prompt trapAsked to invent detail

Confidence is a tone — verify before you act

Ask: What happens if this is wrong?

  • Low risk — Joke captions, silly rename ideas, brainstorming party themes. Skim is enough; wrong is mostly awkward.
  • Medium risk — Homework facts, sports scores for a blog, club meeting times. Check before you publish or submit.
  • High risk — Health, safety, legal trouble, bullying advice, anything that could hurt someone. Do not trust AI alone. Ask a trusted adult, coach, counselor, or official source.

Health and safety posts deserve hard verification: official guidance, a parent, school nurse — not a confident paragraph from a chatbot. Joke captions deserve light verification: does it still sound kind, and did it avoid punching down?

Read

Spend attention like money

You only have so much focus after school. If you verify everything at maximum intensity, you’ll burn out and stop verifying at all. Risk-based checking is how adults stay safe without freezing.

Quick rules:

  1. If it tells you to put something in your body, stop.
  2. If it tells you to meet someone or share a location, stop.
  3. If it only affects a caption tone, keep moving after a kindness check.
  4. If it affects a grade, open a real source.

Read

Design your personal thresholds

Write three lines in your notes app:

  • Low: I skim for kindness and obvious nonsense.
  • Medium: I check one independent source before I submit or post.
  • High: I do not act on AI alone; I ask a trusted adult or official channel.

Then practice sorting for a week. A rumor about a canceled dance is medium (check the school account). A joke about pineapple pizza is low. Advice about ignoring concussion symptoms is high — stop immediately.

Risk-based verification is not laziness. It is how you keep verifying at all. Emergency-level energy on every caption creates burnout; burnout creates blind trust. Match the brake pressure to the road.

Read

Social feed twist

AI-shaped text shows up in comments and “tips.” Apply the same ladder there. If a post tells kids what to swallow, where to meet, or how to hide injuries from coaches, treat it as high risk even if it is written in a friendly voice.

Go deeper

Before you start

Why this matters

Two posts hit your feed. Post A: a joke caption for a meme about cafeteria pizza. Post B: a tip that says “If you feel dizzy after practice, take these leftover pills from a friend — AI says it’s fine.” Same app, wildly different risk. Light laugh vs health and safety.

In the wild

See how this idea shows up as a product and a company — then come back to the lesson. Skills transfer across vendors.

Check your understanding

Page assessment

Answer from memory. Completion is saved from this evidence, not from opening the next page.

1. What question starts a risk-based verification habit?
2. Why should health tips get harder checks than joke captions?
3. Give one medium-risk school example and its check.
4. What is the danger of trying to “fully verify” every tiny AI output?
5. Who are trusted humans you can go to for high-risk questions?

All responses are required.