Chapter 13 Exercises: Governing AI — Policy, Regulation, and Global Approaches

Conceptual Questions

Exercise 13.1: The Four Challenges of AI Governance

The chapter identifies four structural challenges that make AI governance difficult: the pacing problem, the knowledge gap, the jurisdictional problem, and the definition problem. Choose one of these challenges and explain, in your own words, why it is particularly hard to solve. Then propose one creative approach — it does not need to be perfect — that might partially address the challenge you chose.

Exercise 13.2: Risk Classification Practice

For each of the following AI systems, classify it using the EU AI Act's risk framework (unacceptable, high, limited, or minimal risk). Justify your classification by explaining what level of harm the system could cause and to whom.

a) An AI system that generates music playlists based on your listening history b) An AI system used by a university to rank applicants for admission c) An AI chatbot that provides customer service for an online retailer d) An AI system that predicts which employees are most likely to quit and flags them for retention efforts e) An AI system that generates realistic images of public figures saying things they never said f) An AI system that monitors factory equipment to predict maintenance needs

Exercise 13.3: The Pacing Problem in Action

Find a recent news story (from the past year) about an AI development that existing regulations were not designed to address. Describe the development, explain which regulatory gap it exposes, and suggest how governance frameworks might need to adapt.

Exercise 13.4: Values in Governance

The chapter argues that different governance approaches reflect different values. Complete the following table by identifying the primary value or values each approach prioritizes:

Then answer: Which values do you think should be prioritized in AI governance? Why?

Applied Exercises

Exercise 13.5: AI Governance Simulation (Group, 6–10 students)

Your group is a newly formed national AI governance commission tasked with developing a regulatory framework for AI in hiring. You must address the following questions:

1. Should AI hiring tools be regulated? If so, how strictly?
2. Should companies be required to disclose when AI is used in hiring decisions?
3. Should applicants have the right to request human review of AI-generated decisions?
4. Should there be mandatory bias audits? If so, who pays for them and who conducts them?
5. What penalties should exist for violations?

Divide into stakeholder groups: - Employers/HR companies (want flexibility and efficiency) - Job applicants (want fairness and transparency) - AI developers (want clear rules they can comply with) - Civil rights organizations (want strong protections against discrimination) - Small business representatives (want affordable compliance)

Each group prepares a 3-minute position statement. After hearing all positions, negotiate a compromise framework that all parties can accept. Document the final framework and note where compromises were made.

Exercise 13.6: Algorithmic Impact Assessment (Individual)

Choose an AI system that you interact with regularly (a social media algorithm, a recommendation system, a search engine, a voice assistant, etc.) and conduct a simplified algorithmic impact assessment:

1. System description: What does the AI system do? What decisions does it make or influence?
2. Data inputs: What data does it use? Where does that data come from?
3. Affected populations: Who is affected by the system's outputs? Are some groups more affected than others?
4. Potential harms: What could go wrong? Consider accuracy errors, biases, privacy violations, and second-order effects.
5. Existing safeguards: What protections are currently in place (if any)?
6. Recommended improvements: What additional safeguards would you recommend?

Present your assessment in a 2-page written report or a 5-minute presentation.

Exercise 13.7: Comparative Regulation Letter

You are an advisor to a country that is developing its first AI governance framework. The country wants to learn from existing approaches before creating its own. Write a 500-word briefing memo that:

1. Summarizes the key features of the EU, U.S., and Chinese approaches (one paragraph each)
2. Identifies the strengths and weaknesses of each
3. Recommends which elements the country should adopt and which it should avoid
4. Explains how the recommended approach reflects the country's values (you may specify the country's context — e.g., developing economy, strong democratic institutions, large tech sector, etc.)

Exercise 13.8: Self-Regulation Stress Test

Choose one major AI company and locate its published AI principles, responsible AI framework, or ethical guidelines. Then:

1. Identify three specific commitments the company has made.
2. Search for news reports about whether the company has lived up to each commitment.
3. Assess: Is the company's self-regulation credible? What evidence supports or undermines its commitments?
4. Propose one change that would make the company's self-regulation more credible (e.g., external oversight, public reporting, independent audits).

Critical Analysis Exercises

Exercise 13.9: The Regulatory Capture Problem

The chapter discusses "regulatory capture" — the risk that regulated industries gain undue influence over their regulators. Research one historical example of regulatory capture in any industry (not necessarily AI), and then:

1. Describe how the capture occurred.
2. Identify what mechanisms failed to prevent it.
3. Draw lessons for AI governance: what features of the AI industry make it particularly susceptible to regulatory capture, and what safeguards might help prevent it?

Exercise 13.10: ContentGuard Governance Framework (Extended Analysis)

ContentGuard, our anchor example of a content moderation AI, is currently subject to minimal regulation in the United States (due to Section 230 protections) and would face transparency obligations under the EU AI Act.

Design a comprehensive governance framework for content moderation AI systems. Your framework should address:

1. Transparency: What information should platforms be required to disclose about their content moderation algorithms?
2. Accountability: Who is responsible when content moderation AI makes errors — suppresses legitimate speech or fails to remove harmful content?
3. Due process: What rights should users have to appeal content moderation decisions? What role should human reviewers play?
4. Bias: How should regulators assess whether content moderation algorithms treat different viewpoints, languages, or communities fairly?
5. Enforcement: Who should enforce these requirements, and what penalties should apply for violations?

Address at least one objection that platforms would raise against your framework.

Exercise 13.11: The Definition Problem in Practice

The chapter notes that defining "AI" has major regulatory consequences. Consider the following systems and decide whether each should be covered by AI regulation. Explain your reasoning.

a) A spreadsheet formula that automatically calculates a loan applicant's creditworthiness based on their income, debt, and credit score b) An email spam filter that uses machine learning to classify incoming messages c) A self-driving car that uses deep learning to navigate roads d) A weather forecasting model that uses statistical methods to predict temperature e) A chatbot that uses a large language model to answer customer questions f) A database query that automatically flags patients whose vital signs exceed certain thresholds

Where would you draw the line between "AI systems that should be regulated" and "software tools that should not"? Is there a principled way to draw this line, or will it always be somewhat arbitrary?

Reflection Exercise

Exercise 13.12: Your AI Governance Position

Write a one-page position paper answering the question: How should AI be governed?

Your paper should: - State your position clearly (e.g., "I believe AI should be governed primarily through [approach] because...") - Acknowledge at least one legitimate counterargument and respond to it - Reference at least two specific concepts from this chapter - Be written for a general audience (no jargon without explanation)

There is no single right answer. The goal is to develop and articulate a thoughtful, informed position that you can defend with evidence and reasoning.