Algorithmic Bias and Fairness

Active learning works because algorithmic bias is a human problem disguised as a technical one. Students need to trace data flows, weigh trade-offs, and experience the gap between a clean algorithm and messy reality. Case studies, audits, and debates put abstract concepts into real systems where students can see bias emerge, measure its effects, and judge possible fixes.

• 11th Grade
• Computer Science
• 4 activities
• 25 min–40 min

Learning Objectives

1. Analyze how specific features in training data, such as zip codes, can act as proxies for protected attributes like race or socioeconomic status.
2. Evaluate the societal impact of biased AI systems by comparing outcomes for different demographic groups in scenarios like loan applications or predictive policing.
3. Design a mitigation strategy to address bias in a hypothetical machine learning model, detailing steps for data preprocessing or model adjustment.
4. Explain the ethical implications of deploying AI systems that perpetuate or amplify existing societal inequalities.

Case Study Analysis

⏱ 40 min·Small Groups

Case Study Analysis: COMPAS and Hiring Algorithms

Assign small groups one of two documented bias cases (COMPAS criminal risk scoring or Amazon's hiring algorithm). Groups read a summary, identify where bias entered the system, and present findings to the class using a structured claim-evidence-reasoning format.

Prepare & details

Analyze how human biases can be inadvertently encoded into AI algorithms.

Facilitation Tip: In Case Study Analysis, assign roles so each student traces a different entry point for bias in the COMPAS system.

Setup: Groups at tables with case materials

Materials: Case study packet (3-5 pages), Analysis framework worksheet, Presentation template

• Analyze
• Evaluate
• Create
• Decision-Making
• Self-Management

Structured Academic Controversy

⏱ 35 min·Pairs

Structured Academic Controversy: Should Biased AI Be Banned?

Pairs argue that a specific biased AI system should be banned outright, then switch and argue for regulation instead of prohibition. After both rounds, partners synthesize a position that addresses both the harms and the practical tradeoffs of each response.

Prepare & details

Explain the societal impact of biased AI systems in areas like hiring or criminal justice.

Facilitation Tip: For the Structured Academic Controversy, require students to cite specific lines from the case studies when stating their positions.

Setup: Pairs of desks facing each other

Materials: Position briefs (both sides), Note-taking template, Consensus statement template

• Analyze
• Evaluate
• Create
• Social Awareness
• Relationship Skills

Socratic Seminar

⏱ 30 min·Small Groups

Dataset Audit: Find the Bias

Provide groups with a simplified synthetic dataset (e.g., fictional loan approval records). Groups use frequency counts and comparison tables to identify which features correlate with protected characteristics, then present what mitigation steps they would take.

Prepare & details

Design strategies to identify and mitigate bias in machine learning models.

Facilitation Tip: When running the Dataset Audit, display correlation tables on the board and ask students to explain each value in plain language.

Setup: Chairs arranged in two concentric circles

Materials: Discussion question/prompt (projected), Observation rubric for outer circle

• Analyze
• Evaluate
• Create
• Social Awareness
• Relationship Skills

Gallery Walk

⏱ 25 min·Whole Class

Gallery Walk: Mitigation Strategies

Post six posters around the room, each describing a different bias mitigation technique (e.g., re-sampling, fairness constraints, post-hoc correction). Students rotate, evaluate each strategy's strengths and limitations on sticky notes, and the class debriefs on which strategies address root causes vs. symptoms.

Prepare & details

Analyze how human biases can be inadvertently encoded into AI algorithms.

Facilitation Tip: During the Gallery Walk, have students rotate with sticky notes to add questions or suggestions to each mitigation poster.

Setup: Wall space or tables arranged around room perimeter

Materials: Large paper/poster boards, Markers, Sticky notes for feedback

• Understand
• Apply
• Analyze
• Create
• Relationship Skills
• Social Awareness

Teaching This Topic

Teachers should frame algorithmic bias as a design failure, not a data failure. Start with concrete artifacts—code notebooks, dataset columns, court rulings—so students confront the messiness early. Avoid rushing to solutions; instead, model how to hold complexity by asking: ‘Which stakeholders’ values are embedded here?’ Use structured controversies to normalize disagreement and peer review to surface blind spots in students’ own reasoning.

What to Expect

Successful learning shows when students can trace how bias travels from historical inequities into data, identify proxy variables in actual datasets, and articulate why technical fixes alone fail without policy and design changes. They should also propose mitigation strategies that balance fairness with system goals and defend their choices in debate.

Watch Out for These Misconceptions

Common Misconception

During Case Study Analysis, some students may think that if the COMPAS algorithm doesn’t include race as an input, it cannot be biased.

What to Teach Instead

During Case Study Analysis, ask students to examine the dataset columns for proxy variables like prior offenses, neighborhood income, or school attended. Have them calculate correlation scores to show how these variables indirectly encode race, then discuss how engineers could remap or remove these proxies.

Common Misconception

During Dataset Audit, students may assume that collecting more data will automatically reduce bias.

What to Teach Instead

During Dataset Audit, direct students to examine the COMPAS dataset size versus its demographic skew. Use the table to show how adding more data from a biased system amplifies historical inequities, then have them propose data collection changes that target underrepresented groups explicitly.

Common Misconception

During Gallery Walk, students might believe that bias only affects high-stakes domains like criminal justice.

What to Teach Instead

During Gallery Walk, include examples from content recommendation and targeted advertising. Ask students to examine how popularity metrics in recommendation systems can reinforce stereotypes, and have them propose domain-specific fairness metrics to evaluate these systems.

Assessment Ideas

Discussion Prompt

After Case Study Analysis, present students with a biased AI in college admissions. Ask them to identify two ways bias could have entered the system, discuss the consequences for underrepresented applicants, and propose one step an engineer could take to address it.

Quick Check

During Dataset Audit, provide a short description of a job candidate recommender. Ask students to write down one source of bias in the training data, one proxy variable that could lead to unfair outcomes, and one fairness metric to evaluate the system.

Exit Ticket

After the Gallery Walk, ask students to write one real-world example of algorithmic bias discussed in class, one reason why eliminating bias is challenging, and one question they still have about AI fairness.

Extensions & Scaffolding

• Challenge: Ask students to design a fairness metric for a new domain, such as college admissions, and write a one-page justification for its use.
• Scaffolding: Provide a partially completed correlation table for the dataset audit with guiding questions like ‘Which columns correlate with gender?’ to help students start.
• Deeper exploration: Invite a local data scientist or policy maker to discuss how their organization audits for bias before deployment.

Key Vocabulary

Algorithmic Bias	Systematic and repeatable errors in a computer system that create unfair outcomes, such as privileging one arbitrary group of users over others.
Training Data	The dataset used to train a machine learning model; biases present in this data can be learned and amplified by the model.
Proxy Variable	A variable that is correlated with a sensitive attribute (like race or gender) and can inadvertently introduce bias into a model even if the sensitive attribute itself is not used.
Fairness Metrics	Quantitative measures used to assess whether an AI model's outcomes are equitable across different demographic groups.
Disparate Impact	A situation where a policy or practice has a disproportionately negative effect on members of a protected group, even if the policy is neutral on its face.