AI & Digital Ethics Grade 11: AI and Democracy Elections Persuasion Surveillance

Study Guide: AI and Democracy – Elections, Persuasion, Surveillance Grade 11 | Social Studies & Technology Ethics

1. The Driving Question

What happens when the people who decide what you see online—what news you trust, what ads persuade you, even what facts feel real—aren’t just humans with biases, but algorithms trained on your past behavior? And if those algorithms can predict how you’ll vote before you even know yourself, is an election still free and fair?

2. The Core Idea – Built, Not Listed

Imagine you’re running for student body president at Lincoln High. Your campaign team doesn’t just put up posters—they build a digital twin of every voter in the school. They know which TikTok trends make sophomores laugh, which Instagram influencers juniors trust, and which Reddit threads seniors get angry about. Then, they use that data to show each group exactly what they want to hear: for the debate team, a promise to fund new trophies; for the environmental club, a plan to ban plastic water bottles; for the gamers, a pledge to upgrade the computer lab. The messages aren’t lies, but they’re micro-targeted—so precise that no two voters see the same version of "the truth." Meanwhile, a rival campaign hires a bot army to flood group chats with fake polls, making it seem like you’re losing support. By Election Day, half the school isn’t sure what you actually stand for.

This isn’t just a high school problem. In real elections, AI-powered tools do the same thing at scale: recommender systems (like YouTube’s "Up Next" algorithm) amplify content that keeps you watching, even if it’s polarizing; predictive models (like Cambridge Analytica’s psychographic profiling) guess your personality from your likes to craft messages that feel personal; and generative AI (like deepfake videos) can make candidates say things they never did. The goal isn’t always to change your mind—it’s to confirm your biases so strongly that you’ll share the content yourself, turning you into an unpaid campaign volunteer. The result? Elections where the winner isn’t the best leader, but the best at gaming the algorithm.

Key Vocabulary: - Micro-targeting Definition: The practice of using data about individuals (e.g., browsing history, location, social media activity) to deliver tailored political messages designed to influence their vote. Example: In 2016, the Trump campaign used Facebook ads to show anti-Hillary Clinton memes to voters in Michigan who had previously engaged with conservative pages, while showing pro-Trump job-creation ads to undecided voters in Pennsylvania. College-level shift: In political science, micro-targeting is studied as part of "computational propaganda," where the focus expands to include state-sponsored disinformation (e.g., Russia’s IRA troll farms) and the long-term erosion of shared reality.

• Algorithmic amplification Definition: When AI systems (like social media feeds) prioritize content that generates strong emotional reactions—anger, fear, or outrage—because it increases user engagement (and ad revenue). Example: After the 2020 U.S. election, YouTube’s algorithm recommended videos claiming voter fraud to users who had watched conservative news, even though those claims were debunked. The more users watched, the more extreme the recommendations became. College-level shift: Media studies scholars debate whether amplification is a bug or a feature of platform design, linking it to "attention economies" where human psychology is monetized.

• Surveillance capitalism Definition: An economic system where companies (like Meta or Google) collect vast amounts of personal data to predict and influence behavior, often without users’ informed consent. Example: In 2018, it was revealed that the app This Is Your Digital Life (used by Cambridge Analytica) harvested data from 87 million Facebook users, not just the 270,000 who downloaded it. The data was then used to build psychological profiles for political ads. College-level shift: Legal scholars argue that surveillance capitalism challenges democratic norms by creating "asymmetric power" between corporations and citizens, with implications for free will and autonomy.

• Generative AI (in disinformation) Definition: AI tools (like DALL-E or Midjourney) that can create realistic text, images, or videos that never existed, often used to spread false narratives. Example: In 2023, a fake AI-generated image of Pope Francis wearing a puffy white jacket went viral, fooling millions into believing it was real. The same technology can be used to fabricate speeches, protests, or "evidence" of corruption. College-level shift: Ethicists warn that generative AI blurs the line between "truth" and "plausible fiction," requiring new frameworks for media literacy and legal accountability (e.g., "deepfake laws").

3. Assessment Translation

How this appears on assessments: - AP U.S. Government & Politics / AP Computer Science Principles: Free-response questions (FRQs) asking you to analyze a scenario (e.g., a leaked internal memo from a social media company) and evaluate its impact on democratic processes. Rubrics prioritize: - Evidence: Citing specific examples (e.g., Cambridge Analytica, 2016 Russian interference). - Trade-offs: Weighing benefits (e.g., increased voter engagement) against risks (e.g., erosion of trust). - Solutions: Proposing policy or technological fixes (e.g., algorithmic transparency laws, digital literacy education). - SAT/ACT (Reading/Writing): Passages about AI and democracy, with questions testing your ability to: - Identify the author’s argument (e.g., "Does the author believe micro-targeting is a threat to democracy, or a tool for better representation?"). - Compare perspectives (e.g., "How would a tech CEO’s view of data privacy differ from a civil rights lawyer’s?"). - Evaluate evidence (e.g., "Which detail from the passage best supports the claim that social media algorithms polarize voters?"). - State standardized tests (e.g., NY Regents, CAASPP): Short-answer questions like: "Explain one way AI tools could be used to manipulate public opinion during an election. Use an example from the text or your own knowledge." - Proficient response: Names a specific tool (e.g., deepfake videos) and explains how it works (e.g., "A candidate could use AI to create a fake video of their opponent making a racist remark, then spread it on social media to turn voters against them"). - Developing response: Vague (e.g., "AI can make fake news") or lacks an example.

Model Proficient Response (AP FRQ): Prompt: "Some argue that AI-driven micro-targeting in elections is a form of free speech, while others claim it undermines democracy. Using your knowledge of U.S. government and technology, write an argument supporting one of these positions. Include at least one example." Response: Micro-targeting using AI threatens democracy because it replaces public debate with private manipulation. In a healthy democracy, voters should engage with a shared set of facts and arguments—like in a town hall where everyone hears the same candidate speeches. But when campaigns use AI to show different voters different messages (e.g., the Trump campaign’s 2016 Facebook ads), it fragments the electorate into isolated "information bubbles." For example, a voter in Ohio might see an ad promising to bring back coal jobs, while a voter in California sees an ad about climate action—even though the candidate has no plan to do either. This makes it impossible for voters to hold leaders accountable, because no one knows what the candidate actually stands for. While free speech protects the right to persuade, AI-driven micro-targeting turns persuasion into a weaponized, secretive process that prioritizes winning over truth.

4. Mistake Taxonomy

Mistake 1: Overgeneralizing the impact of AI - Prompt: "How does AI influence elections?" - Common wrong response: "AI makes elections unfair because it spreads fake news." - Why it loses credit: Too vague—doesn’t specify how AI spreads disinformation or which tools are used. "Fake news" is a buzzword, not an analysis. - Correct approach: 1. Name a specific AI tool (e.g., generative AI, micro-targeting, bot networks). 2. Explain its mechanism (e.g., "Generative AI can create deepfake videos that make candidates appear to say things they didn’t"). 3. Give an example (e.g., "In 2024, AI-generated robocalls mimicked Joe Biden’s voice to tell New Hampshire voters not to participate in the primary"). 4. Connect to democracy (e.g., "This erodes trust in elections by making voters question whether any information is real").

Mistake 2: Ignoring trade-offs in solutions - Prompt: "Should governments regulate AI in elections? Why or why not?" - Common wrong response: "Yes, because AI is dangerous and needs to be controlled." - Why it loses credit: No consideration of counterarguments (e.g., free speech, innovation) or unintended consequences (e.g., government overreach). - Correct approach: 1. State your position (e.g., "Governments should regulate AI in elections, but carefully"). 2. Acknowledge trade-offs (e.g., "Regulation could limit free speech if it’s too broad, but no regulation allows manipulation"). 3. Propose a specific solution (e.g., "Laws could require social media platforms to label AI-generated content and disclose ad targeting data"). 4. Use evidence (e.g., "The EU’s AI Act is an example of balancing innovation and protection").

Mistake 3: Confusing correlation with causation - Prompt: "Does social media cause political polarization?" - Common wrong response: "Yes, because people argue more online now." - Why it loses credit: Assumes social media is the cause of polarization without considering other factors (e.g., gerrymandering, economic inequality) or the role of algorithms in amplifying existing divisions. - Correct approach: 1. Define polarization (e.g., "Polarization is when people’s political views become more extreme and hostile toward the other side"). 2. Explain how AI contributes (e.g., "Algorithms prioritize content that triggers outrage, which rewards extreme posts and punishes nuanced ones"). 3. Acknowledge other factors (e.g., "Polarization also stems from media echo chambers and political leaders using divisive rhetoric"). 4. Use data (e.g., "A 2021 MIT study found that false news spreads 6x faster on Twitter than true news, often because it’s more emotionally charged").

5. Connection Layer

1. Within subject (Social Studies)-Media Literacy Understanding how AI shapes elections clarifies why media literacy isn’t just about spotting "fake news"—it’s about recognizing how platforms design your attention to serve their goals (e.g., engagement over truth).

2. Across subjects (Computer Science)-Bias in Algorithms The same AI tools used for micro-targeting in elections are used in hiring, lending, and policing. If an algorithm learns from biased data (e.g., historical hiring patterns that favor men), it will replicate that bias—just like how political ads might reinforce racial or gender stereotypes.

3. Outside school-Your group chats Next time you see a viral political meme or a "leaked" video in your Snapchat group, ask: Is this real? Who made it? Why was it sent to me? AI tools make it easier than ever to create and spread persuasive content, turning every share into a potential act of unpaid campaigning.

6. The Stretch Question

If an AI system could predict with 99% accuracy how you’ll vote based on your social media activity, should that data be used to tailor political ads to you—or should it be illegal, even if it means campaigns are less "efficient"?

Pointer toward the answer: This isn’t just about privacy—it’s about whether democracy requires shared information. If every voter sees a different version of a candidate’s platform, how can we have a collective conversation about what’s best for society? On the other hand, if campaigns can’t use data to reach voters, might they rely even more on expensive TV ads that favor wealthy candidates? The tension here is between individualization (treating voters as unique) and collectivism (treating voters as a public with shared interests). Historically, democracies have erred on the side of transparency (e.g., public debates, equal airtime laws), but AI challenges those norms by making personalization too effective. The answer might lie in regulated transparency—forcing campaigns to disclose how they use data, without banning it outright.