When AI Gets It Wrong: Teaching Kids to Catch Hallucinations

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-parenting-when-AI-gets-it-wrong-adults

1. What is an AI hallucination?

   1. When an AI generates creative but unrelated content
   2. When an AI refuses to respond due to safety concerns
   3. When an AI admits it does not have enough information to answer
   4. When an AI confidently states false information as if it were true

2. Why do AI models often sound confident even when their outputs are incorrect?

   1. They are trained to generate fluent, authoritative-sounding text rather than to verify accuracy
   2. They are designed to hedge their responses with uncertainty markers
   3. They have access to real-time databases that sometimes contain errors
   4. They deliberately mislead users for competitive reasons

3. Which type of AI hallucination is generally the MOST difficult to detect?

   1. Obviously fabricated statistics that can be checked with basic math
   2. Confidently stated false information in domains where you have no expertise
   3. Errors in everyday common knowledge that most people know
   4. Greetings or casual conversation responses that are inappropriate

4. A family is discussing AI outputs and one member asks 'How would I know if this were wrong?' before trusting a claim. What habit does this represent?

   1. Source avoidance, where you refuse to look at any external information
   2. Confirmation bias, where you look for evidence that supports what you already believe
   3. Automatic trust, where you assume AI is accurate unless proven otherwise
   4. Active skepticism, where you consider the limits of your own knowledge before accepting information

5. What does the lesson mean by 'confidence calibration' in the context of AI?

   1. Teaching AI systems to express doubt when they are uncertain
   2. The process of matching your own confidence level to the actual reliability of information
   3. A feature that allows users to adjust how confident an AI sounds
   4. A method for measuring how often AI produces accurate outputs

6. Which domain is identified in the lesson as one where AI commonly produces hallucinations?

   1. Personal information about public figures
   2. Basic arithmetic calculations
   3. Simple greeting responses
   4. Weather information for major cities

7. What is a primary source in the context of fact-checking AI claims?

   1. A peer-reviewed article published in the last year
   2. Any source that agrees with what the AI has stated
   3. Original documents, firsthand accounts, or direct evidence related to a claim
   4. The first website that appears in search engine results

8. The lesson suggests making 'check the source' a family mantra. What is the intended purpose of this practice?

   1. To encourage children to distrust their own judgment
   2. To teach children to only trust official-looking sources
   3. To make children skeptical of all information from any source
   4. To create an automatic habit of verification before accepting important claims

9. Why does the lesson recommend showing kids specific examples where AI confidently states something false?

   1. To make children stop using AI entirely
   2. To build recognition and skepticism through real examples
   3. To prove that AI is fundamentally untrustworthy
   4. To demonstrate that AI errors are obvious and easy to spot

10. What does the lesson say about making children skeptical of EVERY AI output?

    1. It would be exhausting and unhelpful
    2. It is necessary to protect children from misinformation
    3. It should be the primary goal of AI literacy education
    4. It would be helpful for building critical thinking skills

11. A family is doing one of the exercises from the lesson. They give an AI a prompt about a historical event and examine the output for likely false elements. What should they look for?

    1. Responses that are too short to be accurate
    2. Claims that contradict well-established historical facts or lack supporting evidence
    3. Obvious spelling errors that indicate the AI is malfunctioning
    4. Content that uses complex vocabulary, which usually indicates errors

12. The lesson mentions five domains for practicing hallucination detection exercises. Which of the following is NOT one of them?

    1. Entertainment news about fictional characters
    2. Science
    3. History
    4. Current events

13. What does the lesson say is the broader goal beyond just catching AI errors?

    1. Replacing traditional education with AI-assisted learning
    2. Building general critical thinking and verification habits that last a lifetime
    3. Making children expert fact-checkers
    4. Teaching children to distrust all technology

14. Why does the lesson recommend involving the whole family in these exercises?

    1. To make learning about AI hallucinations feel like a fun game rather than work
    2. To build shared language and reinforcement of verification habits across household
    3. To ensure children do not use AI without adult supervision
    4. To compete against each other to find the most errors

15. What distinguishes a useful exercise in this lesson from simply being skeptical of AI?

    1. Exercises focus on verifiable domains and provide verification methods
    2. Exercises are meant to be done alone, not with family
    3. Exercises only work with the most advanced AI models
    4. Exercises require expensive tools or subscriptions