Lesson Report:
Title
– Selling Products vs. Selling Ideas: AI, Targeted Advertising, and Democratic Guardrails
– Synopsis: This session transitioned from mapping the AI technology landscape to analyzing how the same infrastructures that optimize commercial advertising are repurposed to influence elections. Through discussion, a whiteboard framework, and a jigsaw reading of the Haleva Amir article on Israel’s 2019–2020 elections, the class examined how microtargeting, data harvesting, and platform affordances can undermine free and fair elections.
Attendance
– Number of students mentioned absent: 0
Topics Covered (chronological)
1) Setup and housekeeping
– Instructor taught from a physical classroom via webcam; audio/video check.
– Reading shift flagged: today’s text was a peer-reviewed academic article (Haleva Amir on Israel’s 2019–2020 elections), different in tone from recent book chapters.
– Video reflection journals: acknowledged receipt of some; deadline reiterated (submit by Sunday); non-scripted submissions acceptable.
– Access: non-AUCA students should have received eCourse credentials; report lingering access issues to the instructor.
2) Framing the week’s pivot: from AI landscape to AI’s impact on democracy
– Prior two weeks: surveyed contemporary AI (especially LLMs) and platform algorithms.
– This week’s focus: how these technologies affect democratic processes.
3) Why do AI systems exist? Class discussion to surface core purposes
– Market replacement/expansion: LLMs (e.g., ChatGPT, Gemini) reduce search friction by delivering natural-language answers, displacing traditional search behavior.
– IT productivity: AI tools accelerate coding and data analysis (e.g., GitHub Copilot); perceived reduction of junior-level programming tasks.
– Data collection: beyond LLMs, platform algorithms (YouTube, Instagram) continuously profile users; student noted privacy and “secret informationâ€� concerns.
4) From commerce to political persuasion: targeted advertising logic
– Mini-lecture: contrasted low-yield billboard ads with high-yield, data-driven targeting on platforms.
– Key concept introduced: targeted advertising as the business model of platform AI, maximizing return on ad spend via behavioral prediction.
– Surveillance capitalism frame (Zuboff): platforms monetize predictions by extracting behavioral surplus (user signals beyond what’s needed for the service).
5) Whiteboard framework: Product → Goal → Behavioral Surplus → Customer
– Commercial example (guided practice):
– Scenario: Kyrgyz-made sneakers buying Instagram ads.
– Goal: click-through → purchase.
– Behavioral surplus: clicks, likes, location, browsing signals feeding back into platform models and ad delivery optimization.
– Customer: the shoe company (pays the platform).
– Political example (transfer of framework to elections):
– Product: targeted ad about the opponent.
– Class-derived goal: not just “change perceptionsâ€� but drive the measurable outcome—votes for our candidate (and/or vote suppression for the opponent) on election day.
– Behavioral surplus: political beliefs/leanings, policy preferences, engagement signals (likes/shares/comments), demographics (age, gender, location, workplace), all used to profile and predict voting behavior.
– Customer: the candidate/campaign (pays the platform/consultants).
– Instructor synthesis: identical infrastructures (commercial ad tech) can be repurposed to sell ideas/candidates and to achieve political outcomes.
6) Democratic guardrails: defining “free and fair� and identifying mechanisms
– Brainstormed components of free and fair elections:
– Universal suffrage and age eligibility (above 18 can vote).
– Periodic elections (regularly scheduled).
– Election security and integrity (every vote counted; no cheating).
– No coercion or manipulation; voters free from undue pressure.
– Financial transparency (who funds, how funds are used).
– Campaign finance limits (caps on donations/expenditures to prevent outsized influence).
– Contextual link to Israel 2019–2020 (per Haleva Amir): security/access largely intact for citizens, but guardrails like campaign finance and anti-manipulation norms were weakened or sidestepped via digital tactics.
7) Jigsaw reading activity: Four mechanisms that undermined Israel’s 2019–2020 elections (Haleva Amir)
– Instructions:
– Four breakout rooms; each group reviewed a specific facet in the article.
– Task: explain how your facet impeded a free and fair election.
– Time: ~10 minutes in groups, then report back.
– Room 1: Fake news and doctored media
– Key points: Deliberate dissemination of false claims and manipulated content, including doctored video targeting Benny Gantz (mis-transcribed as “Benny Grahamâ€�) and election-day disinformation urging voters not to support his party.
– Mechanism: Social accounts (often unattributed) enable plausible deniability; politicians and operatives can amplify misleading memes without formal attribution to the campaign.
– Electoral impact: Erodes trust in media/institutions, misleads voters at critical moments, and bypasses accountability checks that exist for official campaign communications.
– Room 2: Political chatbots and astroturfing
– Key points: Deployment of “Bibi-botâ€� style chatbots to simulate direct interaction with Netanyahu, collect personal data, and microtarget; astroturfing creates the illusion of broad grassroots support via coordinated inauthentic activity.
– Mechanism: Automated personas and orchestrated networks manipulate perceived consensus and tailor messages to psychological and demographic profiles.
– Electoral impact: Distorts the information environment, shifts social norms via perceived majority views, and influences voter behavior without transparency.
– Room 3: Illegal data accumulation
– Key points: Voter data gathered via quizzes, tools, and conversational interfaces without proper consent or legal basis; integrated into profiling pipelines that inform message targeting.
– Mechanism: Circumvents lawful data collection standards and election data regulations; merges personal and political data for prediction/targeting.
– Electoral impact: Unfair advantage to campaigns willing to violate data protections; undermines privacy and fairness principles embedded in campaigning laws.
– Room 4: Anonymous messaging and noncompliance with election law
– Key points: Mass anonymous texts/DMs (e.g., via SMS/WhatsApp) disseminated without attribution; widespread noncompliance with disclosure, funding, and messaging rules; weak enforcement capacity by election authorities and platforms.
– Mechanism: Hard-to-trace messaging channels evade transparency and oversight; traditional regulators under-equipped to monitor/act at platform speed and scale.
– Electoral impact: Voters cannot evaluate source credibility; rules meant to ensure fairness (disclosures, spending caps, source identification) are effectively neutralized.
– Instructor synthesis:
– The same ad-tech stack that makes product sales efficient (profiling, microtargeting, optimization loops) can be used to “sellâ€� ideas/candidates while evading guardrails via anonymity, automation, and data laundering.
– Plausible deniability plus enforcement gaps create incentives to violate norms with limited risk.
8) Closing and look-ahead
– Next class (Thursday): Compare political advertisements across time—legacy broadcast-era tactics vs. contemporary microtargeted campaigns; note how technological affordances change strategy and accountability.
– Homework reminders:
– If you have not yet read the Haleva Amir article, complete it.
– If you have not posted your video reflection journal, submit ASAP (deadline Sunday).
Actionable Items
Immediate (before Sunday)
– Send reminder to students who have not submitted the video reflection journal; reiterate “no scriptsâ€� guideline and deadline.
– Continue reviewing/listening to submitted video reflections; provide brief feedback rubric or acknowledgment to confirm receipt.
High priority (this week)
– Verify all non-AUCA students can access eCourse; follow up with any students reporting login issues; offer a temporary alternative (email or shared folder) for materials and submissions if needed.
– Prepare Thursday’s comparative ad materials:
– Curate 2–3 legacy TV/radio ads and 2–3 modern microtargeted/social examples (include at least one from the Israel 2019–2020 cycle if feasible).
– Draft prompts that guide students to analyze disclosure, targeting, accountability, and measurable outcomes.
– Plan a short activity linking “product/goal/behavioral surplus/customerâ€� to each example.
– Close the loop on today’s whiteboard: post a clean summary image or typed outline of the Product → Goal → Behavioral Surplus → Customer framework and the four Israel mechanisms for students’ notes.
Short-term follow-ups
– Clarify citations in the syllabus/reading list:
– Correct the reference to Papacharissi (ensure accurate author/title).
– Provide full citation for Haleva Amir article for students’ bibliographies.
– Consider a brief primer for next class:
– Definitions sheet (astroturfing, behavioral surplus, microtargeting, plausible deniability).
– One-slide overview of Israel’s election regulatory context (campaign finance limits, disclosure rules, enforcement bodies).
Longer-term (module planning)
– Plan a session on regulatory responses: platform policies, election commissions, data protection laws (e.g., consent standards), and their enforcement challenges.
– Design an assessment where students map a real campaign’s digital tactics to guardrails and propose concrete policy or platform interventions.
Homework Instructions:
ASSIGNMENT #1: Complete the Haleva Amir reading
You will finish reading the Haleva Amir article on the 2019–2020 Israeli elections to deepen your understanding of how AI-enabled targeted persuasion can weaken democratic guardrails. This connects directly to our shift from “AI selling products� to “AI selling ideas/power,� and prepares you for Thursday’s comparison of political advertising across time.
Instructions:
1) Locate the Haleva Amir article assigned for this week (the case study on the 2019–2020 Israeli elections).
2) Read or finish reading the article carefully. As you read, focus on the four facets we worked with in class:
– Fake news and doctored media
– Political chatbots and astroturfing
– Illegal data accumulation
– Anonymous messaging and noncompliance with election laws
3) For each facet, jot brief notes (3–5 bullets) answering:
– What happened?
– How did this tactic undermine a “free and fairâ€� election (think: guardrails like transparency, finance limits, no manipulation/pressure, integrity/security)?
– What data or behaviors were platforms extracting (behavioral surplus) and how did that enable targeting?
4) Make at least one connection to our class framework:
– How targeted advertising shifts from selling products to selling ideas/candidates
– Ties to surveillance capitalism/data collection we discussed earlier (e.g., recommendation algorithms, behavioral profiling)
5) Prepare 3–4 concise points you can share on Thursday when we compare older campaign tactics to today’s AI-enabled tactics.
6) If the academic style is tough, skim the abstract and conclusion first, then read the intro and the section that matches your breakout group’s facet, and finally the remaining sections.
7) Complete this before our next class (Thursday).
ASSIGNMENT #2: Video Reflection Journal (final reminder)
You will submit your brief, unscripted video reflection to synthesize your takeaways from the last two weeks (AI landscape, algorithms beyond LLMs, targeted advertising) and connect them to our current focus on democracy and elections. This gives you space to articulate your perspective in your own voice.
Instructions:
1) If you already submitted your video reflection, you’re all set—no resubmission needed.
2) If you have not submitted yet, record a short, unscripted video reflection. Speak naturally (do not read a script). As noted in class: “As long as you sent something and you weren’t reading a script, you’re going to do just fine.�
3) Suggested prompts to guide you (choose 2–3):
– One key insight about how AI systems and algorithms (e.g., feeds, recommendation engines, LLMs) shape user behavior.
– One connection between targeted advertising for products and targeted persuasion in politics.
– One takeaway from the Haleva Amir case about why elections can fail the “free and fairâ€� test in the digital era.
– One question or concern you still have about AI and democracy.
4) Keep it brief and clear. Identify yourself at the start, make sure audio is audible, and speak conversationally (notes are fine, just don’t read verbatim).
5) Upload your video (or paste an unlisted link) to the assignment page labeled “Video Reflection Journal.�
6) Deadline: Sunday by 11:59 PM (local time).
7) If you recently received your course-site credentials and still have trouble accessing the submission page, contact the instructor immediately so access can be fixed.