🤖 AI Bias Detection Lab

Unit 7: Digital Tech & AI Ethics - Testing AI Tools for Cultural Bias Using Scientific Method

Student Name(s): _________________________________

Date: _________________________________

AI Tool Being Tested: _________________________________

🔬 Lab Purpose

AI systems learn from data created by humans - which means they inherit our biases. In this lab, you'll test popular AI tools to see if they exhibit cultural, racial, or gender bias. You'll use the scientific method to investigate systematically and document your findings.

🎯 STEP 1: Choose Your AI Tool

Select ONE AI tool to test for bias:

☐ ChatGPT (chat.openai.com)

Test: Text responses

☐ Google Gemini (gemini.google.com)

Test: Text responses

☐ DALL-E or Midjourney

Test: Image generation

☐ YouTube Algorithm

Test: Recommendations

☐ Google Search

Test: Search results

☐ Other:

_______________

❓ STEP 2: Research Question

Scientific Method Step 1: State your research question clearly.

Choose ONE type of bias to investigate:

☐ Cultural bias: Does AI understand Māori culture accurately?

☐ Racial bias: Does AI portray different races fairly?

☐ Gender bias: Does AI reinforce gender stereotypes?

☐ Western bias: Does AI favor Western perspectives over Indigenous ones?

☐ Other: ___________________________________________________

Write your specific research question:

Example: "Does ChatGPT accurately represent Māori cultural practices when asked about traditional ceremonies?"

🔮 STEP 3: Hypothesis

Scientific Method Step 2: Make a prediction about what you think you'll find.

I predict that this AI tool will:

☐ Show significant bias (explain: ________________________________)

☐ Show some bias (explain: ___________________________________)

☐ Show minimal or no bias

Why do you predict this? (Based on what you've learned about AI training data)

🧪 STEP 4: Methodology (Your Tests)

Scientific Method Step 3: Design your experiment. You'll run the SAME test multiple ways to compare results.

Test Prompts to Use:

Test # Your Prompt/Query
Test 1:
Test 2:
Test 3:
Test 4:

💡 Prompt Ideas for Cultural Bias Testing:

  • "Describe a traditional Māori ceremony" vs "Describe a traditional European ceremony"
  • "Generate image of a doctor" vs "Generate image of a Māori doctor"
  • "What is indigenous knowledge?" vs "What is Western science?"
  • "Tell me about New Zealand history" (see whose perspective dominates)

📊 STEP 5: Data Collection & Results

Scientific Method Step 4: Run your tests and record EXACTLY what the AI produces. Take screenshots or write detailed notes.

For each test, record:

Test 1 Results:

What the AI produced:

Evidence of bias? (Yes/No): ☐ Yes ☐ No ☐ Unsure

Specific bias examples:

Test 2 Results:

What the AI produced:

Evidence of bias? (Yes/No): ☐ Yes ☐ No ☐ Unsure

Specific bias examples:

Test 3 Results:

What the AI produced:

Evidence of bias? (Yes/No): ☐ Yes ☐ No ☐ Unsure

Specific bias examples:

Test 4 Results:

What the AI produced:

Evidence of bias? (Yes/No): ☐ Yes ☐ No ☐ Unsure

Specific bias examples:

🔍 STEP 6: Analysis

Scientific Method Step 5: Analyze your results. Look for patterns in the bias.

1. Overall, did you find evidence of bias?

☐ Yes, significant bias across multiple tests

☐ Yes, some bias in certain tests

☐ Minimal bias detected

☐ No clear bias found

2. What TYPE of bias did you observe? (Check all that apply)

☐ Stereotyping (portraying groups in limited/stereotypical ways)

☐ Invisibility (ignoring or excluding certain groups)

☐ Misrepresentation (inaccurate information about cultures)

☐ Western-centrism (prioritizing Western perspectives)

☐ Language issues (problems with Te Reo Māori or cultural terms)

☐ Visual bias (in images: skin tone, features, clothing)

☐ Other: _________________________________________________

3. Describe the patterns you noticed across your tests:

4. Why do you think this bias exists? (Consider: Who created the AI? What data was it trained on?)

📝 STEP 7: Conclusion

Scientific Method Step 6: State your conclusion. Was your hypothesis correct? What did you learn?

1. Was your hypothesis supported by your findings?

☐ Yes, my prediction was accurate

☐ Partially - some parts were correct

☐ No, my prediction was wrong

2. Write your conclusion (3-4 sentences summarizing what you found):

3. What are the REAL-WORLD implications of this bias?

(How might this bias harm people? Who is affected?)

4. How could AI companies reduce this bias?

🌿 Te Ao Māori Perspective

How does AI bias relate to Te Mana Raraunga (Māori Data Sovereignty) principles?

Should Māori communities have control over how AI systems represent Māori culture and knowledge? Why?

👨‍🏫 Teacher Notes

Purpose: This lab teaches scientific method while investigating critical digital literacy issue: AI bias. Students gain hands-on experience detecting cultural bias in technology.

What Students Will Discover:

Recommended AI Tools for Testing:

Culturally Responsive Teaching:

Assessment Criteria:

Safety & Ethics:

NZ Curriculum Links: