Chapter 29: Exercises — AI and Democratic Processes

Individual Reflection Exercises

Exercise 1: Your Algorithmic Information Diet For one week, consciously track the political and civic content you encounter via social media or news aggregators. For each piece of content, note: platform encountered on, emotional response it triggered, whether it confirmed or challenged your existing beliefs, and whether you shared it. At the end of the week, write a 500-word analysis of your own information diet and the algorithmic role in shaping it.

Exercise 2: Deepfake Literacy Assessment Without using detection tools, evaluate five audio or video clips of political content online and rate your confidence that each is authentic. Then research each clip and determine whether your assessments were accurate. Write a reflection on your ability to detect AI-generated political content and what this implies about citizen deepfake literacy generally.

Exercise 3: Filter Bubble Test Using an account you believe is in a political "bubble" (or constructing one deliberately), map the political content that recommendation algorithms surface to you. Then deliberately follow or engage with five sources of opposing political content and document how the recommendations change. What does this reveal about filter bubble permeability?

Exercise 4: Content Moderation Dilemma Read three specific content moderation decisions by a major platform — cases where either content was removed that arguably should not have been, or content was left up that arguably should have been removed. Write a 600-word analysis of the decision-making challenges, the role of AI in moderation, and what accountability for those decisions should look like.

Exercise 5: Civic AI Audit Research how AI is being used in your national or local government's administration — in benefits determination, tax enforcement, public safety, or regulatory compliance. Document: what system is used, what decisions it influences, what transparency is provided to affected citizens, and what recourse exists for challenge. Write an assessment of whether these uses meet democratic accountability standards.

Group Discussion Exercises

Exercise 6: The Content Moderation Simulation Divide into groups of 5. Each group is a content moderation team for a hypothetical social media platform facing an election. You have 30 minutes to set policies for: AI-generated political content, coordinated inauthentic behavior, election disinformation, and political micro-targeting. Present your policies and defend them against critique from other groups.

Exercise 7: The Platform Redesign Challenge You are a product team tasked with redesigning Facebook's news feed algorithm to reduce its contribution to political polarization while maintaining business viability. The constraint: the platform's advertising revenue depends on engagement metrics. Develop a redesign proposal and present the trade-offs.

Exercise 8: The Regulatory Design Workshop Design a federal AI and elections regulatory framework for the United States. Your framework must address: AI-generated content in political advertising, algorithmic amplification by platforms, deepfake political content, and micro-targeting of voters. You must navigate First Amendment constraints on political speech regulation. Present your framework to the class for critique.

Exercise 9: The Disinformation Response Exercise You are a campaign communications team. Two days before a major election, an AI-generated audio clip circulating on social media depicts your candidate apparently conceding the election and discouraging supporters from voting. The clip is sophisticated enough that many users believe it is authentic. You have 2 hours to develop a response strategy. Present and defend your approach.

Exercise 10: The Myanmar Policy Review If you were a policy director at a major social media platform in 2015, what policies would you have implemented — given the knowledge available at the time — to reduce the harm that occurred in Myanmar? What organizational structure would have surfaced the warning signals faster? What resources would have been required? Present a specific pre-crisis policy proposal.

Analytical Exercises

Exercise 11: Algorithmic Amplification Analysis Using publicly available data on viral political content from a major platform (Twitter/X academic research API, or published research datasets), analyze the characteristics of content that achieved high sharing velocity. What emotional characteristics dominate? What accuracy level is associated with high virality? Write a 1,000-word analysis of the relationship between engagement optimization and content quality.

Exercise 12: Gerrymandering Analysis Using the Redistricting Data Hub or similar public resources, examine a recently drawn congressional district map in a state of your choice. Assess the degree of partisan optimization using standard efficiency gap or other metrics. Research what AI tools, if any, were used in the redistricting process. Write an assessment of the relationship between algorithmic capability and the partisan map outcome.

Exercise 13: DSA Compliance Assessment Select a major social media platform and assess its compliance with the EU Digital Services Act's systemic risk assessment requirements as they relate to election integrity. Based on the platform's public transparency reports, audit reports, and press coverage, evaluate whether it has: identified relevant risks, implemented proportionate mitigation measures, provided researcher access, and taken specific actions in recent election contexts. Produce a compliance assessment with gap analysis.

Exercise 14: vTaiwan Deep Dive Research the vTaiwan deliberative process and the Polis tool it uses in detail. Analyze one specific policy issue on which vTaiwan was applied. Document: the process, the participation, the AI role, the outcome, and the implementation. Then evaluate: what are the preconditions for this approach to work? Could it be applied in your national context? What would need to change?

Exercise 15: Cambridge Analytica Methodology Review Critically analyze the Cambridge Analytica methodology — specifically the OCEAN personality model and its claimed ability to enable effective voter persuasion. Draw on academic literature on the OCEAN model's predictive validity for political behavior. Evaluate whether the actual capabilities matched the company's marketing claims. Write a skeptical assessment of what the evidence actually shows.

Case Application Exercises

Exercise 16: The Robocall Investigation You are a regulator investigating the New Hampshire Biden robocall. Develop the investigative and enforcement strategy: what laws apply, what evidence do you need, who bears liability (the consultant who ordered it, the vendor who produced it, the distribution service), and what penalties are proportionate. Then draft the regulatory guidance you would issue to address this type of incident prospectively.

Exercise 17: Platform Policy for the 2026 Election You are the Head of Elections Integrity Policy at a major social media platform preparing for the 2026 US midterm elections. Draft a comprehensive election integrity policy addressing: AI-generated content in political advertising, deepfake content of political candidates, coordinated inauthentic behavior, platform recommendations during the election period, and transparency reporting. Your policy must be implementable, enforceable, and defensible against First Amendment challenges.

Exercise 18: AI Democracy Risk Assessment Choose an upcoming election in any country. Conduct a pre-election AI democracy risk assessment covering: disinformation capabilities likely to be deployed, platform weaknesses in the relevant language context, institutional resilience of the media environment, regulatory frameworks available for response, and civil society capacity to monitor and counter AI-generated disinformation. Produce a risk assessment document with specific recommendations.

Exercise 19: The Deliberative Democracy Design Design an AI-assisted democratic deliberation process for a specific policy question in your municipality or country — a transport planning decision, a budget priority question, or a land use controversy. Drawing on the vTaiwan model and other examples, specify: the participation mechanism, the AI tools used, the governance structure, the protection against manipulation, and the relationship between deliberation outcomes and formal decision-making. Present your design.

Exercise 20: Deepfake Legislation Drafting Draft specific legislative language for a federal deepfake-in-elections statute that: prohibits specified forms of AI-generated political content without disclosure, requires platforms to implement disclosure mechanisms for synthetic content, creates enforcement mechanisms and penalties, and survives anticipated First Amendment challenges. Compare your draft to existing state legislation and identify where trade-offs were necessary.

Research and Writing Exercises

Exercise 21: International Comparison of Election AI Governance Compare the approaches to AI and elections governance in three different countries — drawing from examples including the EU (DSA/AI Act), United States, India, Brazil, UK, and South Korea. For each, analyze: what laws and regulations apply, what enforcement mechanisms exist, what specific incidents have been addressed, and how effective the framework has been. Synthesize into a comparative assessment with transferable lessons.

Exercise 22: The Evidence on Echo Chambers Conduct a literature review of academic research on filter bubbles and echo chambers published since 2015. Identify the major empirical studies, their methodologies, and their findings. Where do studies agree? Where do they disagree? What methodological factors explain the discrepancies? Produce a 1,500-word evidence summary that accurately represents the state of academic knowledge.

Exercise 23: AI for Democracy — Opportunity Mapping AI's threats to democracy receive extensive attention; its opportunities receive less. Map the documented uses of AI to strengthen democratic participation — multilingual translation, deliberation platforms, policy analysis tools, accessibility tools for voters with disabilities, and others. For each, assess evidence of effectiveness, scalability, and risk of misuse. Produce a balanced assessment of AI as a tool for democratic strengthening.

Exercise 24: Platform Transparency Report Analysis Obtain and analyze the most recent transparency reports from Meta, Google, Twitter/X, and TikTok related to election integrity enforcement. For each platform, document: what was removed, on what basis, in what volume, through what process, and with what outcomes. Compare across platforms. Write an assessment of the adequacy of transparency reporting for democratic accountability purposes.

Exercise 25: Journalism and AI Disinformation Interview a working journalist at a news organization about their experience with AI-generated disinformation — how they encounter it, how they verify content, what tools they use for detection, and how it has changed their practice. Write a 1,000-word account of the journalist's perspective and what it reveals about the institutional role of journalism in democratic resilience against AI disinformation.