Why AI Can Be Unfair Without Meaning To

End-of-lesson check

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

1. Why might an AI image maker show mostly men when you ask it to draw a CEO?

   1. The AI wants to upset people who see the pictures
   2. The AI learned from pictures that mostly showed men as CEOs
   3. The computer part called the GPU only understands male pictures
   4. The AI randomly picks men every time

2. What is the word that describes when AI copies unfair ideas from the world around it?

   1. Coding
   2. Learning
   3. Honesty
   4. Bias

3. An AI keeps using girl names for nurses and boy names for doctors. Where do you think it learned this?

   1. From counting how many letters are in each job name
   2. From a special computer game about hospitals
   3. From books and articles written by humans over many years
   4. From flipping a coin every time it needs a name

4. If an AI image maker shows you something that feels unfair or wrong, what should you do?

   1. Share it with all your friends without thinking
   2. Ask the AI to show you more unfair images
   3. Push back and tell someone it's not right
   4. Believe the AI because it's always correct

5. Why do engineers (the people who build AI) work to fix bias when they find it?

   1. They want to make the AI more confusing
   2. They want AI to treat everyone equally and fairly
   3. They want the AI to run faster on computers
   4. They want to add more biased ideas on purpose

6. What is 'training data' for an AI system?

   1. The special tests that engineers give AI to see if it works
   2. The instructions that tell a computer when to turn on and off
   3. The examples and information an AI studies to learn how to answer questions
   4. The pictures and videos that people upload to share online

7. An AI text generator always describes chefs as men and nannies as women. This is an example of:

   1. The AI trying to be funny
   2. Bias from learning old ideas about jobs
   3. A mistake where the AI forgot all words
   4. A computer virus affecting the AI

8. Can AI be unfair without meaning to be?

   1. No, AI can only do exactly what programmers tell it
   2. No, computers never make mistakes
   3. Yes, because it learns from a world that already has unfair ideas
   4. Yes, but only if the computer has a broken part

9. The lesson says 'AI can be unfair because the world it learned from was unfair.' What does this mean?

   1. AI only learns from perfect, completely fair sources
   2. AI purposely learns unfair ideas to cause problems
   3. AI copies the unfair ideas that already exist in human writing and pictures
   4. AI reads newspapers that are 100 years old

10. You ask an AI to describe 'a hero' and it always describes a strong man. What could you try next?

    1. Give up and never use AI again
    2. Ask the same question ten more times exactly the same way
    3. Ask again using different words like 'a heroic nurse' or 'a heroic teacher'
    4. Tell the AI it is a bad computer

11. What job did the lesson say AI might incorrectly show as always being done by women?

    1. Nanny or childcare worker
    2. Police officer
    3. CEO or business leader
    4. Pilot or airplane driver

12. Which of these is the BEST reason to question an AI that shows something unfair?

    1. Because AI learns from humans and humans can have unfair ideas
    2. Because AI computers are dangerous and should be turned off
    3. Because AI is always right and you want to test it
    4. Because you want the AI to show more unfair things

13. If AI learned from books written only a long time ago, what unfair ideas might it pick up?

    1. Ideas about how to fix computer bugs
    2. Ideas about which foods taste best
    3. Ideas about how to build faster computers
    4. Ideas that were common long ago but are now considered unfair

14. Why can't an AI just 'decide' to be fair on its own without help from humans?

    1. AI has feelings and doesn't care about fairness
    2. AI only knows what it has learned from human-created data
    3. AI already knows everything about fairness
    4. AI is smarter than all humans and doesn't need help

15. The lesson says engineers work hard to fix bias when they find it. What does this tell us about bias?

    1. Bias is fun and makes AI more interesting
    2. Bias is a problem that people should try to fix
    3. Bias is something computers create on their own
    4. Bias is always caused by broken computer parts