Chapter 16: Exercises

Transparency in AI Marketing and Advertising

25 exercises ranging from analysis to policy design to ethical evaluation.

Part A: Comprehension and Analysis (Exercises 1–8)

Exercise 1: Programmatic Advertising Trace Using publicly available documentation from ad tech companies (Google, Meta, The Trade Desk, or similar), map the data flow in a standard programmatic advertising transaction. Identify: What data about the user is collected? By whom? At what point in the transaction? How is this data used to determine which ad is shown? Who has access to the resulting behavioral profile? Write a 600-word plain-language explanation of this process suitable for a consumer with no advertising industry knowledge.

Exercise 2: FTC Guidelines Application A cosmetics company has hired a social media influencer to promote their products. The company provides the influencer with products for free and pays a flat fee. The influencer posts content about the products without disclosing the commercial relationship. Additionally, the influencer uses an AI writing tool to generate the captions for their posts, which they post without modification or disclosure. Analyze this scenario under current FTC endorsement guidelines: What disclosure requirements apply? Who is legally responsible? What remedies might the FTC seek?

Exercise 3: Proxy Targeting Analysis An employer uses Facebook advertising to recruit for warehouse positions. They target their recruitment ads to users who live within 20 miles of their warehouse facility (in a city where the warehouse district is predominantly Black and Hispanic) and who have expressed interest in "logistics" or "supply chain" content. They use a lookalike audience based on their current workforce (which is 90% white and male). Analyze the discriminatory targeting risks in this scenario. Which targeting parameters raise civil rights concerns? Under which legal frameworks? What would a compliant recruitment advertising strategy look like?

Exercise 4: Dark Patterns Identification Visit three e-commerce websites or app stores and document any dark patterns you encounter. For each dark pattern, describe: the specific design element; how it misleads or manipulates users; which category of dark pattern it represents (subscription trap, confirmshaming, misdirection, etc.); and whether AI personalization might be involved in its implementation. Evaluate whether any of the patterns you found might constitute an unfair or deceptive practice under FTC standards.

Exercise 5: Dynamic Pricing Ethics Research the dynamic pricing practices of two companies in different industries (suggestions: airline, hotel, ride-sharing, e-commerce, insurance). For each company, analyze: What data inputs does the pricing algorithm use? What is the range of price variation across customers? Is there evidence of geographic or demographic pricing disparities? What are the company's disclosure practices regarding dynamic pricing? Apply both legal and ethical frameworks to evaluate whether the pricing practices are acceptable.

Exercise 6: Cambridge Analytica OCEAN Model Research the OCEAN/Big Five personality model and its application in targeted advertising. Write a 700-word critical analysis that addresses: How well does the academic literature support the claim that OCEAN model inferences from social media data predict real-world behavior? What are the methodological limitations of Cambridge Analytica's approach? Does the model's empirical validity affect its ethical status (i.e., is manipulation less ethically problematic if it doesn't actually work)?

Exercise 7: EU AI Act vs. US FTC Disclosure Comparison Compare the AI content disclosure requirements under the EU AI Act (specifically the provisions on limited-risk AI systems and AI-generated content) with the FTC's guidance on AI-generated content and endorsements. For each framework, answer: What must be disclosed? By whom? To whom? In what form? What enforcement mechanisms exist? What gaps remain? Conclude with a recommendation for which framework provides better consumer protection and why.

Exercise 8: Brand Safety AI Audit Research documented cases where brand safety AI has blocked advertising adjacent to content from LGBTQ+, Black, or disability-focused creators. Identify at least three specific documented cases. For each, analyze: What type of content was blocked? What characteristics of the content triggered the AI's classification? What was the impact on creators' revenue? What changes, if any, have platforms made in response? Synthesize your findings into a 500-word assessment of the racial and identity-based bias embedded in brand safety AI systems.

Part B: Applied Design (Exercises 9–17)

Exercise 9: Ethical Advertising Policy Design You are the Chief Marketing Officer of a regional bank that uses AI for all digital advertising, including customer acquisition and product cross-selling. Design a comprehensive ethical AI advertising policy for your bank that addresses: targeting criteria (what will and will not be used); discriminatory targeting prevention, including proactive bias testing; dynamic pricing practices for financial products; AI-generated content disclosure; children's protection; and vendor accountability requirements. The policy should be practical and implementable, not aspirational only.

Exercise 10: Algorithmic Bias Audit Design Design a bias audit protocol for a housing advertising platform's AI delivery optimization system. The audit should address: What demographic disparities would constitute a concern? How would you measure delivery distributions across protected classes? What data would you need, and how would you obtain it? What threshold of disparity would trigger remediation? What remediation options are available? How would you verify that remediation was effective? Specify the statistical methods you would use.

Exercise 11: Meaningful Consent Mechanism Design a consent mechanism for behavioral advertising data collection that meets the standard of "genuine informed consent" — not merely legal compliance. Your design should address: What information must be disclosed for consent to be genuinely informed? How should the consent interface be designed to avoid dark patterns? Should consent be purpose-specific (different consents for different uses of data) or general? How should withdrawal of consent work? Test your design against common dark pattern critiques.

Exercise 12: Deepfake Detection Policy You are the Head of Content Integrity at a major social media platform. Design a policy for detecting and addressing AI-generated synthetic media used in advertising without appropriate disclosure. Your policy should address: technical detection methods; human review processes; advertiser responsibility and consequences; user reporting mechanisms; and coordination with law enforcement for illegal uses of synthetic media. Address the arms race dynamic: how will your policy remain effective as synthesis quality improves?

Exercise 13: Political Advertising Transparency Framework Draft a proposed framework for transparency requirements for AI-powered political advertising that a state legislature might adopt. Your framework should address: what information must be disclosed about the AI targeting parameters used in political ads; who must make disclosures; what constitutes prohibited psychographic targeting; how transparency requirements will be enforced; and how the requirements interact with First Amendment protections for political speech. Explain the constitutional basis for your framework.

Exercise 14: Children's Advertising Protection Design Design a COPPA-compliant system for an educational app that includes some advertising-supported content. Your design should address: how the app will identify users under 13 (and the limitations of this identification); what data collection practices are permissible for users under 13; how ad delivery will differ for verified child users; what parental consent mechanisms apply; and how you will audit compliance. Address the specific technical challenges of age verification in online environments.

Exercise 15: Lookalike Audience Fairness Intervention A retail employer uses a Facebook lookalike audience based on their existing store management workforce (85% white, 70% male) to recruit for management training program. You have been asked to redesign the recruitment advertising strategy to reach a more diverse applicant pool while retaining the efficiency benefits of algorithmic targeting. Design a revised strategy that: reduces the demographic concentration of the lookalike audience approach; uses alternative targeting methods that are less likely to reproduce existing workforce demographics; and includes measurement to assess whether the revised approach is producing more diverse applicant pools.

Exercise 16: AI-Generated Content Disclosure System Design a disclosure labeling system for AI-generated or AI-assisted marketing content. Your system should address: at what threshold of AI involvement disclosure is required (purely AI-generated vs. AI-assisted with substantial human input); how disclosures should be displayed (formatting, placement, language); whether different types of AI involvement require different types of disclosure; and how your system applies to different content types (text, image, video, audio). Draft sample disclosure language for each category.

Exercise 17: Contextual Advertising Business Case Prepare a business case for a mid-sized e-commerce company to transition from behavioral/retargeting advertising to contextual advertising. Your business case should address: the revenue impact of reducing behavioral targeting (based on available research on contextual vs. behavioral advertising effectiveness); the privacy-related costs the company currently bears (GDPR compliance, data breach risk, consumer trust erosion); the reputational benefits of a privacy-first approach; and the implementation costs and timeline of the transition. Include a financial model with explicit assumptions.

Part C: Case Discussion and Role Play (Exercises 18–22)

Exercise 18: Fair Housing Enforcement Simulation Role-play scenario: A fair housing organization has filed a complaint with HUD alleging that a property management company's Facebook recruiting advertising has produced racially disparate delivery — based on algorithmic ad delivery data obtained through the DSA research access provisions. The property management company denies discriminatory intent and argues that the delivery patterns reflect authentic user interest in their properties. Divide into groups representing HUD, the fair housing organization, and the property management company. Each group presents its legal and policy arguments. The class decides what remedy, if any, HUD should require.

Exercise 19: Brand Crisis Management Role-play scenario: A major consumer goods brand discovers that its AI-powered advertising system has been showing predatory payday loan ads (from a co-advertising partner) targeted specifically at users the algorithm has identified as financially distressed. The brand was unaware of this practice; it was implemented by its ad tech vendor without disclosure. A journalist has asked for comment. Divide into roles: the brand's CEO, CMO, General Counsel, and Head of Communications must develop a response strategy. The class evaluates the strategy for adequacy and ethical appropriateness.

Exercise 20: Democratic Ethics Forum Structured debate: The question is whether AI-powered political micro-targeting should be prohibited in federal elections. The affirmative argues that psychographic targeting exploits psychological vulnerabilities, creates information asymmetry that undermines democratic deliberation, and has been used by foreign actors to interfere in elections. The negative argues that political targeting is protected political speech, that the government should not regulate political messaging, and that the answer to bad speech is more speech, not regulation. Each side presents, then faces cross-examination.

Exercise 21: Platform Accountability Role Play Role-play scenario: You are a panel of members of Congress conducting a hearing at which the CEO of a major social media platform is testifying about discriminatory advertising practices. Divide into groups: legislators who will conduct the hearing; the platform's CEO and counsel who will testify; public interest advocates who have submitted written testimony; and advertising industry representatives who have submitted written testimony. The hearing should address: what the platform knew and when; what steps it took to address discriminatory delivery; what additional regulatory requirements are appropriate; and what the platform's responsibility is for its algorithms' discriminatory effects.

Exercise 22: Advertising Ethics Board Role-play scenario: An advertising ethics board reviews three cases submitted by consumer advocates:

Case 1: An insurance company uses AI to identify which customers are most anxious about their financial security (based on browsing patterns indicating financial stress) and targets them with fear-based ads for expensive whole-life insurance products. No specific claims in the ads are false.

Case 2: A political campaign uses AI to identify which voters are most susceptible to negative emotional appeals about immigration based on inferred personality characteristics, and sends these voters targeted ads featuring alarming crime statistics that are technically accurate but highly selective.

Case 3: An alcohol brand uses AI to identify users who have recently expressed emotional distress on social media and targets them with ads for its alcohol products during these periods of vulnerability.

The board must decide whether each practice is: (a) clearly ethical; (b) ethically acceptable but requiring disclosure; (c) ethically problematic but not regulatable; (d) ethically unacceptable and requiring prohibition. Provide reasoning for each decision.

Part D: Research and Extended Analysis (Exercises 23–25)

Exercise 23: Global Comparative Analysis Compare the regulatory approaches to AI in advertising in four jurisdictions: United States, European Union, United Kingdom, and one of (Canada, Australia, India, Brazil). For each jurisdiction, examine: requirements for advertising disclosure; civil rights/anti-discrimination requirements applicable to digital advertising; data collection requirements for behavioral targeting; political advertising regulation; enforcement mechanisms and notable enforcement actions. Write a 2,500-word comparative analysis concluding with your assessment of which jurisdiction has the most effective and comprehensive approach and why.

Exercise 24: Algorithmic Discrimination Research Proposal Design an empirical study to test whether AI advertising delivery systems produce discriminatory outcomes for a specific category (housing ads, employment ads, or credit ads) and a specific platform. Your research design should specify: research question and hypothesis; data collection methodology; measurement of demographic distribution in ad delivery; statistical tests for disparate impact; the platform access or data access your research requires; and how you would handle platform resistance or denial of access. Discuss the ethical considerations in your research design, including potential tensions with platform terms of service.

Exercise 25: Strategic Transformation Plan You have been hired as a consultant to a major digital advertising agency that has decided to make a genuine, not just performative, commitment to ethical AI advertising. The agency's CEO has asked you to design a transformation plan that addresses: (1) auditing existing client campaigns for discriminatory targeting; (2) developing internal standards for advertising AI use across the agency; (3) vendor selection criteria that include ethical advertising AI standards; (4) client education and communication about the transition; (5) measuring and reporting on the agency's ethical advertising performance; and (6) engaging with industry bodies to raise standards across the advertising sector. Write a 2,000-word transformation plan with specific action steps and a 12-month implementation timeline.