AI and hallucination vs mistake: spot when AI is making it up

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-builders-foundations-AI-and-hallucination-vs-mistake-r7a10-teen

1. Why should you verify a quote that an AI attributes to a famous person?

   1. Famous people are in AI training data, so all their quotes are guaranteed accurate
   2. AI models have perfect recall and never make up quotes, so verification is just a formality
   3. Verifying quotes is unnecessary because AI only outputs facts
   4. AI can generate plausible-sounding quotes that the person never actually said

2. Which topic is an AI most likely to hallucinate about?

   1. Historical events from 50 years ago
   2. Simple vocabulary definitions
   3. Very recent events that happened after the AI's training cutoff
   4. Basic arithmetic problems

3. What does the term 'confabulation' mean in AI contexts?

   1. When AI correctly cites its sources
   2. When AI fills in gaps by generating plausible but false information
   3. When AI refuses to answer a question
   4. When AI intentionally lies to deceive users

4. An AI provides you with three academic paper citations. What is the best next step?

   1. Search for the papers online to confirm they actually exist
   2. Cite them immediately in your homework
   3. Copy the citations into another AI to double-check
   4. Assume they are real since AI has access to databases

5. Why does next-token prediction lead to hallucinations?

   1. The model runs too fast
   2. The model is trying to deceive users for fun
   3. The model has a limited vocabulary
   4. It predicts the most likely next word based on patterns, not verified facts, so it can generate believable but false content

6. What type of person is an AI more likely to hallucinate about?

   1. A popular movie character
   2. A famous historical figure everyone knows
   3. A current world leader featured in daily news
   4. A niche academic expert with few online records

7. What is the most important AI literacy skill mentioned in this topic?

   1. Memorizing all AI terminology
   2. Being able to spot when AI is making up facts
   3. Learning to write better prompts
   4. Knowing how to code in Python

8. When AI gives you a confident, specific answer, what should you typically do?

   1. Ask the AI to explain its confidence level
   2. Ignore the answer entirely
   3. Accept it immediately since confidence indicates accuracy
   4. Be skeptical and verify the specific claims

9. Which of these is an example of an AI hallucination?

   1. AI gives a correct answer but with typos
   2. AI says it doesn't know the capital of a small country
   3. AI invents a quote from a historical figure who never said it
   4. AI says 2+2=5

10. What makes hallucinations particularly risky compared to other AI errors?

    1. They use more computer processing power
    2. They are always obvious to spot
    3. They only affect older AI models
    4. They appear confident and factual, making them hard to detect

11. If an AI tells you about a breakthrough scientific discovery from last week, what should you suspect?

    1. The AI is testing you
    2. The AI might be hallucinating recent events
    3. The AI has access to real-time information
    4. The discovery is definitely real

12. A student asks an AI for three sources for a research paper and gets citations. What should the student verify?

    1. Whether each paper actually exists and contains relevant information
    2. Only the author names
    3. Only the publication dates
    4. Nothing if the titles sound professional

13. Why might an AI invent a fake person in a biography it generates?

    1. It fills gaps by generating plausible details when it lacks specific information
    2. It has too much information about obscure people
    3. It is playing a game with the user
    4. It is trying to impress the user

14. What is fact-checking in the context of AI use?

    1. Letting AI check its own answers
    2. Only checking information you already doubt
    3. Trusting AI more than human sources
    4. Manually verifying AI output against reliable external sources

15. Which statement about AI confidence is most accurate?

    1. AI confidence is unrelated to accuracy
    2. Confident AI answers should be verified just like uncertain ones
    3. AI is programmed to always show low confidence
    4. Confident AI answers are always correct