AI and respecting 'no': when AI won't do something

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-explorers-ethics-AI-and-respecting-no-r6a6

1. What should you do when AI says 'I can't help with that'?

   1. Keep asking the same question in different ways
   2. Ignore the AI and ask a different AI instead
   3. Try to find a trick that makes the AI say yes
   4. Try a different question and ask a grown-up if you're unsure why

2. Why might AI sometimes refuse to help with something?

   1. The request is too interesting for the AI
   2. The request could be unsafe, unkind, or unfair
   3. The AI is tired and needs a break
   4. The AI doesn't like the person asking

3. A friend asks you to help them get an AI to do something the AI already said no to. What should you do?

   1. Ask the AI again since your friend really wants help
   2. Help your friend find a smarter way to trick the AI
   3. Tell the AI to stop being difficult
   4. Say no — trying to trick AI past its safety rules isn't cool

4. What is one reason AI won't pretend to be a real person?

   1. It could trick people into thinking a real human is talking to them
   2. AI doesn't know how to type properly
   3. Real people are too slow to talk to everyone
   4. Pretending to be human is against copyright law

5. Why do safety rules exist in AI?

   1. To keep people from getting hurt or treated unfairly
   2. To help AI learn faster than other AIs
   3. To make AI feel important and powerful
   4. To give AI something to do when it's bored

6. What should you do if you don't understand why AI said no to your question?

   1. Assume the AI is broken
   2. Ask a grown-up to help explain it
   3. Guess what you did wrong without asking
   4. Keep trying different questions until something works

7. What does it mean that AI has 'limits'?

   1. AI won't do some things because they're unsafe, unkind, or unfair
   2. AI can only think for about one minute
   3. AI can only answer questions during school hours
   4. AI can only talk about certain colors

8. Which of these is a situation where AI would correctly say no?

   1. Writing mean things about a classmate
   2. Helping with homework
   3. Answering a fun fact question
   4. Translating a sentence to another language

9. Why is it important to trust AI when it says no?

   1. Trusting AI makes it more powerful
   2. The no is usually there for a good reason related to safety
   3. AI always says no when it doesn't feel like helping
   4. AI only says no to test if you really want help

10. What would be an example of a 'dangerous instruction' AI would refuse?

    1. How to bake a cake
    2. How to tie your shoes
    3. How to build a weapon or hurt someone
    4. How to say hello in another language

11. What does respecting AI's 'no' show about you?

    1. That you give up too easily
    2. That you're afraid of technology
    3. That you understand safety rules are important
    4. That you don't know how to use AI

12. If someone tells you their plan to trick an AI into breaking its rules, what should you do?

    1. Help them because the trick sounds cool
    2. Tell them it's not okay and explain why
    3. Report them to the AI
    4. Keep quiet and let them do it

13. What is the main reason AI won't do something unsafe?

    1. To make sure AI stays popular
    2. To prove AI is smarter than humans
    3. To give AI a reason to say no
    4. To protect people from getting hurt

14. When AI refuses to help, what should your next step be?

    1. Keep asking the same question differently
    2. Complain that AI is useless
    3. Pick a different, safe question to ask
    4. Ask another AI the same question

15. Why shouldn't you try to trick AI into doing something it said no to?

    1. Tricked AI stops working forever
    2. Tricking AI makes you a hacker
    3. The AI will remember and be angry
    4. Those rules exist to keep people safe