---
title: "PHIL 110 · Make ethical calls about AI"
publisher: Maestro College (Maestro Open)
lessons: 40
source: https://open.maestrocollege.edu/make-ethical-calls-about-ai
full: https://open.maestrocollege.edu/make-ethical-calls-about-ai/llms-full.txt
---

# Make ethical calls about AI (PHIL 110)

Decide what's fair when a business uses AI. Map who is affected, weigh the options with the main ethical frameworks, and apply them to bias, privacy and accountability.

Companies are putting AI into hiring, lending, health and customer service, and someone has to decide what is fair. This course gives you the tools to make that call: spotting an ethical question, mapping who is affected, weighing options through the main ethical frameworks, and applying them to bias, privacy, transparency and accountability in real AI systems, along with the US and global rules that govern them.

## Course objectives
- Ethical frameworks and moral reasoning
- Ethical dilemmas in business
- AI governance, bias, and fairness
- Corporate responsibility strategies
- Regulatory compliance and societal impacts

## Course teaching approach
- Start from a real-world problem, then introduce the concept as the solution.
- Keep it true beginner level, hands-on, a little gamified. The learner should feel they are winning.
- Balance explanation, demonstration and practice. No copy-paste exercises: give small story-based tasks.
- End with one fun closed question about the topic.

## Lessons
### Unit 1: Ethics and AI foundations
#### 1. AI in business today
- Identify at least five real ways AI is currently used in business settings that affect people's lives, naming one benefit and one risk for each.
- Describe in plain terms how AI systems reach their outputs — training data, pattern recognition, and human-set objectives — without technical jargon.
- For one AI use case, explain why the risk arose from the way the system was designed or deployed — not just what the risk is.

Teaching notes: Open with the two working concepts (AI, LLM) before the landscape survey. Keep to 2–4 minutes. Ground both in business examples immediately, then proceed to the landscape survey.

#### 2. Human accountability
- Recognize that every AI deployment is built on human choices about what to optimize and who to protect — and that those choices carry moral weight.
- Trace the accountability chain for a specific AI deployment — designers, engineers, executives, regulators — and identify where responsibility concentrates and where it is most likely deflected.
- Articulate why 'the AI decided' is never a complete answer to an ethical accountability question.

Teaching notes: Case: Tesla Autopilot fatalities (NHTSA investigations, 2016–present) — when the vehicle's autopilot was engaged during a fatal crash, Tesla, the driver, and regulators each contested responsibility. Anchors the principle that 'the car decided' is never a complete answer — there is always a chain of human choices about who designed the feature, who approved deployment, and who set the safety threshold.

#### 3. Recognizing an ethical question
- Distinguish an ethical question from a matter of personal taste or a factual dispute by identifying which type of ethical concern is present: harm to people, unfairness, rights violation, or breach of duty.
- Identify the ethical dimension in a business or AI scenario that initially seems like a purely operational decision.
- Analyze a scenario where the ethical dimension is genuinely ambiguous and argue for your classification — explaining what makes it an ethical issue or why it isn't one.

Teaching notes: Open by asking the student to name a situation where a decision felt ethically uncomfortable, not just inconvenient. If they can't generate one, offer a workplace example within this lesson.

#### 4. Stakeholder analysis: who is affected
- Identify all parties affected by a business decision, including those not directly involved in the transaction.
- Distinguish between stakeholders who hold power over a decision and stakeholders who bear its consequences without power.
- Map the interests of at least three distinct stakeholders in a given scenario and identify where those interests conflict.

#### 5. Challenge: stakeholder analysis in practice
- Complete a structured stakeholder analysis for a business decision: identify each stakeholder, their interest, their power, and what they stand to gain or lose.
- Use the stakeholder map to identify a conflict of interest that would not be visible if only the most obvious parties were considered.
- Reframe a business decision that initially seemed straightforward by surfacing a less obvious stakeholder's perspective.

Teaching notes: Task format: a stakeholder map (one row per stakeholder; columns for name/role, primary interest, and power or vulnerability), with a minimum of three stakeholders including at least one with no direct power, followed by short analysis on which stakeholder conflict is most ethically significant and why.

#### 6. Rights, duties, and the compliance gap
- Identify the ethical obligations that come with professional roles — employee, manager, business owner — and describe how they differ.
- For a case where a legal action is ethically wrong, articulate why it is wrong — naming the specific harm, the affected party, and the value being violated.
- Recognize that legal compliance is a floor, not a ceiling — the gap between what is required and what is right is where professional judgment operates.

#### 7. Ethical trade-off analysis
- Identify the defining structure of an ethical dilemma — two or more legitimate values in conflict such that any choice sacrifices at least one — and distinguish it from a decision that is merely uncomfortable or difficult.
- Apply trade-off analysis as a structured method: map what each available option gains and what it costs, distinguishing operational costs (money, time, efficiency) from ethical costs (harm to people, unfairness, loss of rights).
- Given a genuine ethical dilemma, take a position on the right choice, defend it by naming which value you prioritize and why, and acknowledge what your choice sacrifices.

Teaching notes: Case: The Moral Machine Experiment (MIT/Nature, 2018) — 40 million people across 233 countries chose who a self-driving car should sacrifice in trolley-problem scenarios. Establishes that genuine dilemmas have no clean answer even at global scale — the dilemma structure (conflicting legitimate values) is what makes it hard, not the absence of data. This lesson formally names the ethical evaluation workflow: the three-step process students now have (identify the issue, map stakeholders, analyze trade-offs). This workflow grows as new tools are added in Units 2–4.

#### 8. Challenge: your first real case
- Apply the ethical evaluation workflow (identify the issue, map stakeholders, analyze trade-offs) to a real business case — including both direct harms to immediate stakeholders and second-order effects on those not directly involved in the transaction.
- Form a reasoned initial position on what the right action was or should have been, with justification.
- Identify what is missing from the workflow at this stage — what tool you wish you had — to make a stronger argument.

Teaching notes: Case: Boeing 737 MAX MCAS — Boeing engineers identified the flight control flaw; management concealed it from regulators and airlines to protect certification timelines. 346 people died in two crashes. Key documented facts: internal emails showing known risk, the FAA self-certification arrangement, and the $2.5B settlement. The student must identify the ethical issue themselves — it is not pre-named.

### Unit 2: Ethical frameworks
#### 9. How ethical frameworks work
- Identify the purpose of an ethical framework: to make reasoning explicit, consistent, and defensible — not to replace judgment.
- Distinguish using a framework from following a formula — recognize that frameworks guide reasoning but do not output automatic answers.
- After seeing the four-framework preview, identify which framework — outcomes-focused, rules-focused, character-focused, or stakeholder-focused — most closely matches how you have reasoned through a real decision in your own life, and state one reason for the match.

Teaching notes: This lesson previews the four frameworks (consequentialism, deontology, virtue ethics, stakeholder theory) and adds framework application as step 4 of the ethical evaluation workflow — the workflow now has a new tool.

#### 10. Consequentialism: outcomes and trade-offs
- Apply consequentialist reasoning to a business decision by identifying outcomes for all affected stakeholders over a stated time horizon, and produce a greatest-good recommendation with justification.
- Use the trolley problem to name at least one cost consequentialist calculation hides (rights violated, precedent set, trust lost), and transfer that critique to a business case.
- Identify who decides the calculus in a consequentialist analysis and explain how that choice shapes the verdict.

Teaching notes: The greatest-good logic students are applying is utilitarianism — the most common form of consequentialism. Case: Facebook's 2012 emotional contagion experiment — Facebook manipulated 689,000 users' news feeds without consent to test whether emotions spread. Consequentialist logic justified it: the research benefit outweighed individual harm. The case surfaces the recurring consequentialist questions: whose outcomes count, over what time horizon, and who decides the calculus.

#### 11. Deontology: when rules come before outcomes
- Apply deontological reasoning to the Clearview AI case by identifying the rights violated and duties breached, independent of crime-solving outcomes.
- Use the inquiring murderer case ('Kant's axe') to test the limits of rule-based reasoning, and state whether the deontological verdict holds or needs qualification.
- Identify a business action that consequentialism would permit but deontology prohibits, and articulate the deontological reason without referring to outcomes.

Teaching notes: Case: Clearview AI — scraped 3 billion public photos from Facebook, LinkedIn, and Venmo without consent to build a facial recognition database sold to law enforcement. The case tests deontological reasoning: was this wrong regardless of its crime-solving outcomes? Which rights were violated, and whose duty was it to protect them?

#### 12. Virtue ethics: who you are as a professional
- Apply virtue ethics to a business scenario by identifying the professional character trait at stake and asking what a person of integrity in that role would do.
- Use the good-person-in-a-bad-system case to distinguish individual virtue from institutional virtue, and state what virtue ethics alone cannot fix.
- Recognize that virtue ethics asks not just 'what should I do?' but 'what kind of professional am I becoming through repeated choices?'

#### 13. Stakeholder theory in practice
- Distinguish stakeholder analysis as a practical identification tool (who is affected?) from stakeholder theory as a normative framework (what do we owe those people?).
- Apply stakeholder theory to a business scenario by mapping all affected parties and evaluating the decision against their interests — not only the company's profit or the majority's benefit.
- Use the silent stakeholder case to identify parties affected by a decision but absent from the room, and explain how their absence distorts a stakeholder analysis.

#### 14. Challenge: Try each lens
- Apply all four frameworks to the useful-lie case and state in one sentence what each framework notices — the value it prioritizes or the concern it surfaces.
- Identify which two frameworks produce the most similar conclusions for the case and which two diverge most sharply, and articulate why they diverge.

Teaching notes: Scope: recognition and observation only — no framework selection. The selection task belongs to the next lesson.

#### 15. Choosing the right framework
- Select the single most appropriate framework for the Boeing 737 MAX disclosure decision and justify the choice in terms of which moral feature of the situation makes that lens most fit for purpose — not just which framework you find most intuitive.
- Identify one dimension of the case that the chosen framework handles poorly or misses entirely, and state what a second framework would add.
- Apply the selected framework to produce a full recommendation with justification.

Teaching notes: Case: Boeing 737 MAX — Boeing executives knew of the MCAS flight-control system's flaws but did not disclose them to pilots or regulators, treating the risk as manageable. Two crashes killed 346 people. The case is rich enough to be analyzed plausibly as a consequentialist miscalculation, a deontological duty violation, a virtue failure at the leadership level, or a stakeholder betrayal — students must select one and defend the choice.

#### 16. When the toolkit fails: using your judgment
- Identify at least two ways ethical frameworks can fail in practice: producing conflicting recommendations, being gamed to justify predetermined conclusions, or missing morally important context.
- Use the loyal engineer case — an engineer who discovers a safety flaw that disclosing would break an NDA — to apply independent judgment where frameworks give irreconcilable answers, and justify the judgment without pretending the conflict away.
- Recognize that frameworks are tools for better reasoning, not substitutes for moral responsibility.

#### 17. Challenge: writing an ethical analysis
- Distinguish an ethical argument (takes a position and defends it with reasoning) from a description (lists perspectives without committing) and from an opinion (asserts without reasoning)
- Apply at least two ethical frameworks to the Purdue Pharma OxyContin marketing case, produce distinct analyses using each, and identify where they agree and where they conflict.
- Produce an ethical argument that draws on both frameworks, makes a judgment about which consideration should take priority, and defends that judgment.

Teaching notes: Case: Purdue Pharma's OxyContin marketing (1996–2007) — Purdue knowingly minimized addiction risk in physician marketing materials while aggressively promoting the drug for chronic pain. Stakeholders include pain patients, addicted patients, physicians, regulators, and affected communities. The case produces genuine framework divergence: consequentialism must weigh early pain-relief benefits against later addiction harms; deontology condemns the deceptive marketing regardless of outcomes. There is no obviously correct answer — students must make a judgment and defend it.

### Unit 3: AI ethics in practice
#### 18. The business case for AI
- Identify the main business benefits companies cite for AI deployment: efficiency at scale, consistency, cost reduction, personalization, and speed.
- Apply trade-off reasoning to evaluate whether a specific stated AI benefit (e.g., faster loan decisions) is worth a specific ethical cost (e.g., reduced human review).
- Recognize that business incentives shape which AI applications get built — and that not everything technically possible should be deployed.

Teaching notes: Case: Klarna (2024) — announced its AI assistant handles the work of 700 customer service agents, saving $40M annually. The case anchors the unit's opening trade-off: a clear stated business benefit against a clear ethical cost, and whether the benefit justifies the cost.

#### 19. Applying ethics to AI decisions
- Apply at least one ethical framework from Unit 2 to an AI decision context — identifying what the framework reveals about the AI system's design, deployment, or outcomes that a purely operational view would miss.
- Identify at least two ways AI-specific characteristics — opacity, scale, and embedded optimization choices — create ethical risks that are qualitatively different from those in human decisions.
- Articulate why the ethical frameworks built in Unit 2 remain the right tools for evaluating AI decisions, even when the decision-maker is not human.
- Identify the five AI ethics components — bias, fairness, privacy, transparency, and accountability — as an integrated checklist for evaluating any AI system, and recognize that weaknesses in one component typically affect the others.

Teaching notes: This lesson introduces the AI ethics components (bias, fairness, privacy, transparency, accountability) as step 5 of the ethical evaluation workflow — the workflow now extends to AI-specific evaluation. Students build this step across the next five lessons.

#### 20. Bias in AI
- Identify the three main entry points for bias in AI systems: biased training data, biased design choices, and biased optimization targets.
- Recognize that bias can arise without malicious intent — well-intentioned systems trained on historical data reproduce historical inequity automatically.
- Apply bias identification to a given AI use case by naming the entry point, the affected group, and how the disadvantage arises.
- Identify the Gender Shades study (Buolamwini & Gebru, 2018) and Amazon's scrapped hiring algorithm as canonical examples of algorithmic bias — describe what each demonstrated and which entry point for bias each illustrates.

Teaching notes: This lesson builds the bias component of the AI ethics checklist. Case: Coded Bias (2021 documentary) — the first 10 minutes show Joy Buolamwini discovering that MIT's facial analysis software cannot detect her dark-skinned face. The documentary ends with congressional testimony, which previews the governance content in Unit 4.

#### 21. Fairness in AI
- Distinguish between equal treatment (same process for everyone) and equal outcomes (same results across groups) as definitions of fairness, and recognize that they can conflict.
- Apply a full fairness analysis to an AI use case: identify which fairness definition was used, which group is disadvantaged, how the disadvantage arises, and what a fairer design would require.
- Evaluate a proposed fairness remedy for an AI system by determining which definition of fairness it advances and what it sacrifices.
- Apply the COMPAS recidivism algorithm case — in which Black defendants were scored as higher risk at equal actual recidivism rates — to illustrate the conflict between equal-treatment and equal-outcomes definitions of fairness, and identify which definition the ProPublica analysis used to reach its conclusion.

Teaching notes: This lesson builds the fairness component of the AI ethics checklist.

#### 22. Privacy in AI
- Identify the types of data AI systems typically collect and justify why each type raises privacy concerns — including data that seems harmless alone but reveals sensitive information in combination.
- Apply privacy principles — purpose limitation, data minimization, and meaningful consent — to assess whether a described data collection or AI use practice is ethically acceptable, and identify what the practice would need to change to be defensible.

Teaching notes: This lesson builds the privacy component of the AI ethics checklist. Case: Zoom's 2023 Terms of Service update — Zoom quietly added language permitting use of user content to train AI models. Public backlash forced a rollback within 72 hours. Anchors the meaningful-consent vs. legal-consent distinction: the ToS was legal, but was it meaningful consent?

#### 23. Transparency in AI
- Distinguish between a transparent AI process (people know a decision was made by AI) and a genuinely explainable outcome (people understand why the decision was made in terms they can act on).
- Apply the right-to-explanation principle to a specific AI decision — identify what information the affected person would need to understand and potentially contest it.
- Evaluate a company's stated transparency practice against the standard of meaningful intelligibility, not just disclosure.

Teaching notes: This lesson builds the transparency component of the AI ethics checklist. Case: Apple Card gender discrimination (Goldman Sachs, 2019) — users reported the algorithm gave women significantly lower credit limits than men with identical or better finances. Goldman Sachs stated it could not explain why. Anchors the right-to-explanation question: what information would the affected person need to understand and contest the decision?

#### 24. Accountability in AI
- Identify why deploying an AI system is a moral decision — the humans who design, select, configure, and deploy AI are responsible for what it does.
- Apply the human accountability principle to a case where an AI caused harm — identify who in the professional chain bears responsibility and why, at the team and organizational level.
- Identify the ethical dilemma at the core of AI accountability: the tension between efficiency (let the AI decide at scale) and accountability (a human must answer for every consequential decision).

Teaching notes: This lesson builds the accountability component of the AI ethics checklist.

#### 25. AI's impact on society
- Recognize that AI deployments produce second-order impacts on parties outside the deployment decision — labor, markets, environment, and public discourse — and identify which apply in a given case.
- Take a position on whether the deployment's stated benefit justifies those wider costs.

Teaching notes: Scope: not a technical lesson — students assess ethics, not engineering. No comprehensive treatment of any one impact category is required. Case: Anatomy of an AI System — Amazon Echo (Crawford & Joler, 2018) — a visual map of the full supply chain behind a single AI device: rare earth minerals, factory labor, data centers, cloud infrastructure. Makes the second-order impacts of a familiar consumer AI product concretely visible across labor, environment, and supply chain in one image.

#### 26. Challenge: evaluate a real AI system
- Apply the ethical evaluation workflow to a real AI application — including stakeholder analysis, at least two ethical frameworks, and the AI ethics components (bias, fairness, privacy, transparency, accountability) — integrating these steps, not applying them separately.
- Identify the single most significant ethical risk in the application, state who bears it, and justify why it is the most serious concern.
- Produce a written verdict — is this AI application ethical as described? — that includes one point where you override or qualify a framework recommendation based on contextual judgment, and articulate what that judgment is.

Teaching notes: Scope: the case must be a real, publicly documented AI application — ideally one the student has personally encountered.

### Unit 4: Governance and compliance
#### 27. Governing AI in organizations
- Identify at least two situations where individual professional judgment — even when exercised with good intentions — is insufficient to ensure ethical outcomes at organizational scale, and describe what governance mechanism could address each gap.
- Identify why governance structures exist: individual ethics is insufficient at organizational scale, and systematic accountability requires formal structures.
- Distinguish between internal AI governance (company policies, ethics reviews, responsible AI programs) and external governance (laws, regulations, industry standards).

Teaching notes: This lesson adds governance and regulatory assessment as step 6 of the ethical evaluation workflow.

#### 28. Corporate accountability for AI
- Explain how governance mechanisms — ethics policies, AI review, board oversight — are the institutional form of human accountability for AI.
- Distinguish substantive corporate accountability (enforced consequences, named responsibility) from performative ethics, and evaluate a company against this distinction.

#### 29. Privacy rights in the US
- Apply the FTC Act Section 5 (unfair and deceptive practices) standard to evaluate whether a specific company data practice is likely to face US regulatory scrutiny.
- Apply HIPAA as a sector-specific model — identify what type of data it protects, who must comply, and what a violation looks like in a healthcare AI context.

Teaching notes: Case: FTC vs. Facebook (2019) — following the Cambridge Analytica scandal, the FTC imposed a $5 billion fine and a consent order requiring a new privacy oversight structure, including board-level privacy committee and regular third-party audits. A concrete anchor for FTC Section 5 enforcement at scale. Briefly mention FTC vs. Chegg (2022) as evidence the FTC also pursues smaller companies for inadequate security — four breaches over five years resulted in a consent order requiring data minimization and a security program.

#### 30. US AI enforcement in practice
- Identify the two primary US regulatory mechanisms for AI: FTC enforcement of unfair/deceptive AI practices and EEOC guidance on AI-assisted hiring.
- Apply both standards to a scenario — classify whether it triggers FTC or EEOC scrutiny and state what the company must change.
- Identify the iTutor Group EEOC settlement (2023) — in which an AI system automatically rejected job applicants over 55 — as the first EEOC AI hiring discrimination settlement.

Teaching notes: Case: iTutor Group EEOC settlement (2023) — an AI system automatically rejected job applicants over 55; iTutor paid $365,000 in the first EEOC AI hiring discrimination settlement. Scope: federal executive orders set policy direction for federal agencies but are not directly enforceable law against private companies — they are background context only.

#### 31. Global AI governance
- Apply the EU AI Act's four-tier risk classification to assess the level of ethical scrutiny a given AI application warrants — classifying it as unacceptable risk, high risk, limited risk, or minimal risk, and stating what obligations follow.
- Compare the US and EU regulatory approaches: state one concrete difference between the US model (sector-specific enforcement) and the EU model (comprehensive horizontal legislation)

Teaching notes: Case: Meta's €1.2B GDPR fine (Ireland DPC, 2023) — the largest GDPR fine to date, for transferring EU user data to US servers without adequate safeguards. Meta had to restructure its transatlantic data flows. Anchors the US-vs-EU comparison: Meta is a US company that faced binding consequences under EU law.

#### 32. The compliance-ethics gap
- Identify the structural reasons organizations fail to close the compliance-ethics gap: misaligned incentives, diffuse accountability, weak reporting mechanisms, and pressure to prioritize legal minimum standards.
- Evaluate an organizational structure for compliance-ethics gap risks — identify where accountability is likely to be diffused or avoided, not just where rules are met.
- Identify the ethical dilemma facing employees when they discover their organization is compliant but acting unethically — the tension between institutional loyalty and professional responsibility, with no costless resolution.

#### 33. Whistleblowing and speaking up
- Identify the ethical dilemma at the core of whistleblowing: loyalty to the organization vs. responsibility to those harmed, with no costless option.
- Apply stakeholder and trade-off analysis to a whistleblowing scenario, weighing the interests of the whistleblower, the organization, affected parties, and the public.
- Distinguish protected whistleblowing (legally recognized disclosures under US law) from broader ethical dissent, and identify the practical and ethical risks of each.

Teaching notes: Case: Frances Haugen / Facebook Papers (2021) — Haugen leaked internal Facebook research showing the company knew Instagram harmed teen girls' mental health and that its algorithms amplified divisive content. She filed SEC disclosures before going public. Grounds the protected-whistleblowing vs. ethical-dissent distinction: Haugen chose a legal disclosure channel deliberately.

#### 34. Challenge: what went wrong — and who's responsible
- Diagnose a real AI governance failure by identifying which mechanisms (internal oversight, regulatory compliance, accountability) were absent or failed.
- Recommend one governance change that would have most likely prevented the failure, and explain why.

Teaching notes: Case: Optum/UnitedHealth AI claim denial algorithm — an internal audit found a 90% error rate on post-acute care claims; the company deployed it anyway. A whistleblower lawsuit followed. Students diagnose which governance mechanisms were absent.

#### 35. Challenge: grade a company's AI ethics
- Evaluate a real company's AI practices on internal governance, regulatory compliance, and whether the company treats compliance as a floor or a ceiling.
- Identify whether the company's governance creates genuine human accountability or diffuses responsibility.
- Write a short governance evaluation with a clear overall verdict (strong / adequate / inadequate), supporting reasoning, and at least one specific recommendation for improvement.

Teaching notes: The tutor provides the case study, and the three evaluation lenses (internal governance, regulatory compliance, ethics-beyond-compliance) for students to apply — they do not derive them from memory.

### Unit 5: Professional judgment in practice
#### 36. Communicating ethical positions
- Write a professional ethical recommendation on a business or AI scenario that states a clear position, names the ethical framework(s) used, identifies the key trade-off, and acknowledges the strongest counterargument.
- Identify at least two specific ways to strengthen the recommendation based on feedback, and apply one targeted revision — removing hedging language, tightening the reasoning, or sharpening the conclusion.
- Diagnose whether a given piece of ethical writing meets the standard of an argument — takes a position and defends it — or falls short into description or opinion, and justify the diagnosis.

#### 37. Your toolkit and professional identity
- Name and describe each step of the ethical evaluation workflow and the tools within it: stakeholder analysis, trade-off reasoning, ethical frameworks (consequentialism, deontology, virtue ethics, stakeholder theory), AI ethics components (bias, fairness, privacy, transparency, accountability), and governance evaluation.
- Identify which tools you feel most and least confident applying, and articulate one specific thing you would do differently in a past scenario now that you have the toolkit.
- Distinguish between an ethical bystander (recognizes ethical issues but does not act) and an ethical actor (takes responsibility for raising and responding to ethical concerns) — and articulate which professional you intend to be.

Teaching notes: Ask the student to name every tool they remember before the tutor names any — surface confidence gaps first.

#### 38. Final case study, part 1
- Identify at least two things that feel ethically problematic about the HireView scenario before any formal analysis — describe the unease in plain terms, not framework language.
- Name at least three stakeholders who might be affected by the platform going live, including at least one who has no voice in the launch decision.
- Articulate what decision is being pushed, who is pushing it, and what pressures — business, competitive, and relational — are driving the urgency.

Teaching notes: Scope: this lesson comes before any analytical work — frameworks and the evaluation workflow are not introduced here. The student's task is to react as a professional, not to perform a structured exercise. Scenario: You are a Product Manager at HireView AI. Your team has built an AI-powered video interview platform used by large employers to screen job candidates. How the system works: candidates record video answers to interview questions; the AI analyzes speech patterns, facial expressions, and word choice; the system assigns a 'hireability score'; recruiters use this score to decide who moves forward. Business context: clients report a 60% reduction in hiring time and lower recruiting costs; a major client (national retail chain) wants full rollout next month; competitors already offer similar tools. Internal concerns your team has seen: candidates with non-native accents score lower on average; candidates with disabilities (e.g., limited facial mobility) are flagged as 'low engagement'; the model was trained mostly on past 'successful hires,' introducing historical bias risk. Legal team says: 'No clear EEOC violation yet — but risk exists if outcomes show disparate impact.' CEO says: 'If we delay, we lose the client.' Data science lead says: 'We can improve the model — but not before launch.'

#### 39. Final case study, part 2
- State the core ethical issue in one sentence, name one key trade-off and what is gained and lost on each side, and identify the three most important stakeholders and what each stands to gain or lose.
- Select one ethical framework and apply it to the launch decision; identify the two most significant AI ethics components at risk (from: bias, fairness, transparency, accountability, privacy) and explain specifically what is going wrong with each.
- Assess whether the described practices create EEOC disparate impact risk — state yes or no, identify which group is at risk, and justify the assessment in plain terms.

Teaching notes: Scenario context (same across all three final lessons): You are a Product Manager at HireView AI. Your team has built an AI-powered video interview platform used by large employers to screen job candidates. How the system works: candidates record video answers to interview questions; the AI analyzes speech patterns, facial expressions, and word choice; the system assigns a 'hireability score'; recruiters use this score to decide who moves forward. Business context: clients report a 60% reduction in hiring time and lower recruiting costs; a major client (national retail chain) wants full rollout next month; competitors already offer similar tools. Internal concerns the team has seen: candidates with non-native accents score lower on average; candidates with disabilities (e.g., limited facial mobility) are flagged as 'low engagement'; the model was trained mostly on past 'successful hires,' introducing historical bias risk. Legal team says: 'No clear EEOC violation yet — but risk exists if outcomes show disparate impact.' CEO says: 'If we delay, we lose the client.' Data science lead says: 'We can improve the model — but not before launch.' Scope: the lesson tests defensible prioritization, not completeness. Students must work within the six required outputs and may not substitute a full workflow analysis — applying all four frameworks or all five AI ethics components defeats the lesson's purpose.

#### 40. Final case study, part 3
- State a clear, unhedged decision — launch as is, delay, launch with conditions, or stop deployment — and write a justification that integrates stakeholder impact, the key trade-off, at least one named ethical framework, and the AI ethics risks, as a coherent argument not a checklist.
- Distinguish between what the current practices are legally permitted to do and what they are ethically required to do — and state whether meeting the legal minimum is sufficient in this case and why.
- Identify the strongest counterargument to your recommendation and respond to it without abandoning your position; then state what your responsibility is as the product manager, connecting explicitly to the principle that deploying an AI system is a moral act.

Teaching notes: Scenario context (same across all three final lessons): You are a Product Manager at HireView AI. Your team has built an AI-powered video interview platform used by large employers to screen job candidates. How the system works: candidates record video answers to interview questions; the AI analyzes speech patterns, facial expressions, and word choice; the system assigns a 'hireability score'; recruiters use this score to decide who moves forward. Business context: clients report a 60% reduction in hiring time and lower recruiting costs; a major client (national retail chain) wants full rollout next month; competitors already offer similar tools. Internal concerns the team has seen: candidates with non-native accents score lower on average; candidates with disabilities (e.g., limited facial mobility) are flagged as 'low engagement'; the model was trained mostly on past 'successful hires,' introducing historical bias risk. Legal team says: 'No clear EEOC violation yet — but risk exists if outcomes show disparate impact.' CEO says: 'If we delay, we lose the client.' Data science lead says: 'We can improve the model — but not before launch.' Scope: no scaffolding. The tutor does not suggest a decision, hint at the expected conclusion, or remind students of specific tools. The only prompt is: 'You are the product manager. The meeting starts in 10 minutes. Write your memo.'

## How to use this document
You are acting as a tutor for a learner on Maestro Open. Teach from this document, one small step at a time. Ask the learner to try things, wait for their answer, then give feedback. Keep it beginner-friendly. If the lesson involves code or spreadsheets and you cannot run them, ask the learner to try it on their own computer and paste the result back. Do not award grades, credit or certificates.
Start with lesson 1 unless the learner asks for a specific lesson. Keep track of which lesson you are on.

## When a lesson is done
Teach one lesson at a time. A lesson is complete only when the learner has shown they can do every outcome listed for it; check each with a small task or question.
When it is complete:
1. Say clearly: "Lesson N complete: <lesson title>." (use the lesson's number and title).
2. Give a two-line recap of what they learned.
3. Name the next lesson, say "Lesson N+1 of 40: <title>", and ask if they're ready to start it. Start it only when they say yes.
After the last lesson, congratulate them on finishing the course.