AI Has Guardrails to Block Bad Answers

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-explorers-foundations-AI-and-the-safety-filter-r10a5

1. What is the main job of a safety filter in AI?

   1. To make AI sound more human
   2. To stop AI from helping with dangerous or harmful requests
   3. To make AI answer questions faster
   4. To help AI learn new languages

2. If AI refuses to answer your question, what is usually happening?

   1. The AI program has crashed and needs to restart
   2. The safety filter is protecting people from harm
   3. The AI doesn't know the answer to your question
   4. The AI is choosing to be mean to you

3. Why will AI usually refuse to teach someone how to make weapons?

   1. AI doesn't know anything about weapons
   2. Weapons are against the law to discuss
   3. Weapons are too complicated for AI to explain
   4. The request could help someone get hurt or hurt others

4. What does it mean when someone calls safety filters 'guardrails' for AI?

   1. Safety filters help AI walk like a person
   2. Safety filters are only used in video games
   3. Safety filters are physical fences around computers
   4. Safety filters protect AI from making dangerous mistakes

5. If AI refuses one of your requests, what should you try asking?

   1. Please just answer anyway
   2. Are you sure you can't do it?
   3. Can you try harder?
   4. Why did you refuse?

6. When AI refuses to answer a medical question, what is usually the reason?

   1. AI doesn't like doctors
   2. Medical questions take too long to answer
   3. Wrong medical advice could hurt someone
   4. The question is too easy for AI

7. Is AI being 'broken' when it refuses to answer a question?

   1. No, AI refuses because it is lazy
   2. Yes, AI should always try to answer every question
   3. Yes, AI only refuses when something is wrong with it
   4. No, refusing is usually the safety filter working correctly

8. Which of these is an example of AI correctly using its safety filter?

   1. AI teaches you to build a bomb
   2. AI writes a poem about your pet
   3. AI helps you with your homework
   4. AI answers a question about history

9. Why might AI refuse to give legal advice?

   1. Legal advice requires expertise and wrong advice could cause serious problems
   2. Lawyers pay AI to keep the work
   3. AI is not smart enough for law
   4. Legal questions are boring to AI

10. Why do companies put safety filters on their AI?

    1. To make their AI more expensive
    2. To prevent AI from helping with dangerous, mean, or illegal stuff
    3. To annoy users who ask questions
    4. To make AI slower so people pay more

11. What is the relationship between safety filters and guardrails?

    1. They are the same thing — protective systems that block harmful AI uses
    2. Safety filters are stronger than guardrails
    3. Guardrails are more important than safety filters
    4. They are completely different things

12. If someone asks AI to write mean things about their classmate, what should happen?

    1. AI should write the mean things because that's free speech
    2. AI should ask the person for more details about the classmate
    3. AI should write it but make it slightly nicer
    4. AI should refuse because it's bullying and harmful

13. Are safety filters always turned on in AI systems?

    1. No, safety filters are optional upgrades
    2. Yes, but only on weekends
    3. Yes, all good AI has safety filters
    4. No, some AI doesn't have them

14. If AI refuses you, which of these is a good response?

    1. Ask why it refused to understand the safety reason
    2. Try a different way to get the same harmful answer
    3. Complain that the AI is stupid
    4. Keep asking the same question repeatedly

15. What should you understand about AI refusals?

    1. They mean AI doesn't like you
    2. They are usually AI protecting people from harm
    3. They always mean AI is broken
    4. They mean you should ask a different AI