Hexagonal Thinking
Is AI Fair? Mapping the Ethics
45 min28 students9th GradeBloom’s Level: Analyze, Evaluate, Create
Before Class
Materials Needed
- ☐Hexagonal Concept Cards (1 set per group)
- ☐Hexagonal Thinking Guide
- ☐Reflection Worksheet
- ☐Case Study Brief
- ☐Scissors and tape
Space Needed
Flat table or floor space for arranging hexagons
Topic Overview
Students will understand how computers learn from examples through supervised and unsupervised learning.
Key Questions
- ?Differentiate between supervised and unsupervised learning paradigms.
- ?Explain the role of training data in supervised learning models.
- ?Predict appropriate applications for each type of machine learning.
About the Methodology
Students receive hexagonal cards with key concepts, people, events, or ideas. They must arrange the hexagons so that touching sides represent connections, then explain why they placed each hexagon where they did. No single "correct" arrangement exists. The value is in the reasoning and discussion.
Learn about this methodologyTime Range
25-40 min
Group Size
12-32
Space Needed
Flat table or floor space for arranging hexagons
Bloom’s Level
Analyze, Evaluate, Create
Fun Factor
Peak Energy Moment
The 'Resumes' scenario feels like a mystery to solve. Students love pointing out the 'stupidity' of a smart machine.
The Surprise
The 'Aha!' moment usually happens when students realize 'Profit' and 'Fairness' are often pulled in opposite directions on their map.
What to Expect
Expect high engagement as they physically move cards and argue over which connection is 'more right.'
Mission Timeline
5m
7m
25m
8m
Spark Briefing Action Debrief
Academic Standards
CSTA: 3A-AP-13
Spark
5 min • Scenario
Read Aloud
A major tech company uses an AI to sort through 10,000 resumes. The AI was trained on the resumes of people currently at the company, most of whom are men. As a result, the AI begins automatically rejecting any resume that contains the word 'women\'s' (like 'women\'s chess club captain') because it hasn't seen that word in 'successful' profiles before. Is this AI broken, or is it doing exactly what it was told to do?
Teacher Notes
Use this to highlight that AI doesn't have a moral compass; it only follows patterns in the data we provide.
Briefing
7 min
Today we use Hexagonal Thinking to map how AI bias works. You will receive 18 hexagonal cards. Your goal is to arrange them so that every touching side represents a logical connection between two concepts. You must be able to explain why two cards are touching.
Group Formation
Divide the class into 7 groups of 4 students each.
Materials Needed
- •Hexagonal Concept Cards (1 set per group)
- •Hexagonal Thinking Guide
- •Reflection Worksheet
- •Case Study Brief
- •Scissors and tape
Action
25 min • 100% Physical
15 min
Cut out the 18 hexagonal cards and read the 'Amazon Hiring Case Study' brief provided in the materials.
Have one student cut while others read the case study to save time.
24 min
Identify the 'Core' of the problem. Place the 'Discrimination' and 'Training Data' cards in the center of your desk.
Walk around and ensure groups aren't just making a straight line; encourage a cluster.
38 min
Connect the remaining 16 cards. For every connection made, the group must agree on a one-sentence justification.
Challenge groups to find 'triple points' where one hexagon touches two others simultaneously.
45 min
Gallery Walk: One 'expert' stays at the desk while other team members rotate to see how different groups connected concepts like 'Profit' and 'Accountability'.
Listen for students debating whether 'Profit' belongs near 'Bias' or 'Regulation'.
53 min
Complete the Reflection Worksheet, documenting your three strongest connections and proposing one 'fix' for the case study.
Collect these as your primary assessment of their conceptual understanding.
If things go sideways
- ▸If a group is stuck, ask: 'How does the lack of Representation in a tech company lead to Bias in the Training Data?'
- ▸If a group is moving too fast, ask them to find a connection between 'Autonomy' and 'Surveillance'.
- ▸Check for groups placing cards randomly; ask them to explain the link between any two touching edges.
Differentiation Tips
- ▸Advanced: Provide three blank hexagons and ask them to add their own complex terms like 'Feedback Loops' or 'Algorithmic Redlining'.
- ▸Struggling: Provide a 'Starter Map' showing the first three connections (e.g., Training Data -> Bias -> Discrimination).
- ▸ELL: Use the visual icons on the cards and allow them to draw a small picture on the back of each card to represent the term in their native language.
Debrief
8 min
Discussion Questions
- 1
Which connection was the hardest to agree on and why?
- 2
If we could only change one hexagon to make AI 'fairer,' which one would it be?
- 3
Who should be held responsible when an algorithm makes a biased decision: the coder, the company, or the data?
Exit Ticket
In your own words, how does 'Training Data' create 'Discrimination' in an AI system?
Connection to Next Lesson
Next mission, we will move from identifying bias to coding a simple decision-tree algorithm that attempts to filter for fairness.
Was this helpful?