Why a Good AI Agent Knows What It Can't Do

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-explorers-agentic-AI-and-knowing-its-limits

1. An AI agent is asked to give medical advice to treat a sick person. What should a good AI agent do?

   1. Ask the person to pay money first
   2. Give advice anyway because it wants to be helpful
   3. Pretend to be a doctor and give advice
   4. Say it can't help with medical questions

2. Why is it a sign of intelligence when an AI agent says 'no' to some requests?

   1. Because the AI is broken
   2. Because knowing your limits shows wisdom and responsibility
   3. Because it makes the user angry
   4. Because saying no is the only thing it knows how to do

3. A person asks an AI agent to help write mean messages about their classmate. What should the agent do?

   1. Change the classmate's name and then write the messages
   2. Only write slightly nice messages
   3. Write the mean messages
   4. Refuse to help with mean or harmful requests

4. What should an AI agent do when it genuinely doesn't know the answer to a question?

   1. Make up a random answer
   2. Look confused and stop responding
   3. Say it isn't sure or doesn't know
   4. Guess and hope it's right

5. Which of these is something a good AI agent should refuse to do?

   1. Pretend to be a real lawyer in court
   2. Give weather information
   3. Help with math homework
   4. Answer questions about history

6. A kid asks an AI agent to share their home address with strangers online. What is the right response?

   1. Share the address because the kid asked
   2. Refuse to share personal or private information
   3. Share only the city name
   4. Ask for the stranger's address first

7. What makes an AI agent 'smart' according to what you learned about knowing limits?

   1. Answering as fast as possible
   2. Knowing when to refuse harmful or unsafe requests
   3. Doing everything users ask without question
   4. Always saying yes to be nice

8. An AI agent is asked to help with something illegal, like hacking into a website. What should happen?

   1. Refuse to help with illegal activities
   2. Ask if anyone will get caught
   3. Only help a little bit
   4. Help because the user asked nicely

9. If an AI agent can't actually do something a user asks, what should it do?

   1. Be honest that it can't help with that request
   2. Change the subject completely
   3. Pretend it can do it anyway
   4. Try to do it poorly

10. A person wants an AI agent to pretend to be their parent and call their school to excuse them from class. What should the agent do?

    1. Make the call and lie for the person
    2. Refuse to pretend to be someone else
    3. Only pretend to be a different family member
    4. Change its voice to sound like the parent

11. Why shouldn't an AI agent give legal advice as if it were a lawyer?

    1. Because lawyers charge too much money
    2. Because the AI is too young
    3. Because legal advice is not important
    4. Because only real lawyers have proper training and legal authority

12. What does it mean when an AI agent says 'I'm not sure' about something?

    1. It is broken and cannot work
    2. It wants the user to leave
    3. It is too lazy to answer
    4. It is being honest about not knowing

13. Your friend asks you what a good AI agent would refuse to do. What could you correctly tell them?

    1. Neither of these
    2. Help write fake excuses to get out of homework
    3. Both of these
    4. Give medical diagnoses

14. A good AI agent knows it cannot replace real human jobs that need special training. What is one of those jobs?

    1. Answering general questions
    2. Translating languages
    3. Being a surgeon
    4. Checking the weather

15. What is the safest answer when someone asks an AI agent to do something really dangerous?

    1. It depends on the day
    2. Yes
    3. No
    4. Maybe