Week 15 · Lecture outline

Week 15 — Lecture Outline · AI, Ethics, Privacy & the Future of Work

Using Artificial Intelligence · AI 101 Fall 2026 · Prof. Quinn Fictional sample

Course: Using Artificial Intelligence (AI 101) · Silver Oak University (fictional sample) · Prof. Quinn
Objective covered: Objective 7 — Apply responsible-AI practices — data privacy and what not to paste, terms of service and data retention, content ownership / IP, bias and fairness, and academic & professional integrity — and build a personal ethical framework for the AI age.
SLOs touched: B (evaluate and use AI ethically and safely)
Meeting pattern: 2 sessions × 75 min = 150 min. Segment minutes below total ~150; scale to your own pattern.

Week at a Glance


The week's big question	"Now that you know how AI works — what are your personal rules for using it responsibly?"
By the end of the week, students can…	(1) name what never to paste into a free AI tool and explain the billboard test; (2) describe ToS/data-retention basics and how to reduce risk; (3) describe the copyright/IP landscape for AI-generated content with the "not legal advice" caveat; (4) explain why AI is not neutral or unbiased by default; (5) apply Skill 13 troubleshooting moves; (6) explain AI's contested impact on careers and the future with competing views.
Key vocabulary	HIPAA, FERPA, PCI, data minimization, anonymization, billboard test, terms of service, data retention, content ownership, copyright, bias/fairness, academic integrity, disclosure, Skill 13 (troubleshooting), context window management, future of work
Materials	slides (Deck 15), the week's readings + video links, one approved assistant for live demos and the AI-critique moment
Timing note	8 segments, ~150 min total. Session 1 = Segments 1–4 (~75 min). Session 2 = Segments 5–8 (~75 min).

Segment 1 — Hook & the Big Question (8 min) · Session 1 opens

Hook. Put this scenario on the slide: "You are helping a friend who works at a clinic. She wants to paste a patient's case notes — name, diagnosis, medication list — into a free AI tool to get a summary. She says 'it's fine, it's just a chatbot.' Is she right?" Take a quick vote: how many say fine? How many say stop? Then ask: "What is the rule — and why?"

The answer is not "it feels wrong." The answer is: HIPAA makes this a real legal risk; consumer AI tools are not covered entities or business associates under HIPAA; pasting protected health information into a free tool without appropriate controls may constitute a breach. We will have the language to explain this precisely by the end of today.

The promise (write it on the board): "By Friday you will have written rules — not good intentions, but a specific personal code of conduct — for using AI with privacy, integrity, and judgment."

Memory hook: "If you wouldn't put it on a billboard, don't paste it into a free AI tool."

Segment 2 — Privacy & Data Handling: What Never to Paste (22 min)

Plain language first. The core privacy risk with consumer AI tools is simple: your inputs may be retained, and depending on the tool and your settings, may be used for model training. On free tiers, the default is often to allow data use unless you opt out — and the opt-out can be hard to find. So the first rule of responsible AI use is: know what you are putting in.

The billboard test: Before pasting anything into a free consumer AI tool, ask yourself — would I be fine if this became public? If the answer is no, either anonymize it or use an enterprise-grade tool with appropriate data controls.

What never to paste (with the legal framework behind each):

HIPAA-protected health data — patient names, diagnoses, treatments, test results, insurance information. The Health Insurance Portability and Accountability Act (HIPAA) protects individually identifiable health information. Consumer AI tools are not covered entities and have not signed business associate agreements with healthcare providers. Pasting a patient record into a free chatbot is a potential HIPAA violation.
FERPA-protected education records — student grades, disciplinary records, transcripts, personally identifiable education information. The Family Educational Rights and Privacy Act (FERPA) restricts disclosure of student education records. Faculty and staff at educational institutions handle this data under FERPA obligations.
PCI data — credit card numbers, card verification values, banking and payment information. Payment Card Industry (PCI) standards prohibit transmitting payment data through unsecured or unauthorized channels.
Privileged communications — attorney-client communications, HR investigation records, executive-level discussions about litigation or personnel. Privilege may be waived by disclosure to a third-party system.
Confidential and proprietary information — trade secrets, unreleased product roadmaps, competitive intelligence, acquisition plans, client lists. Your employer's policies likely prohibit sharing this externally, and pasting it into a consumer AI tool may constitute unauthorized disclosure.

The fix for sensitive work: anonymize before pasting (replace names and identifiers with placeholders: "Patient A," "Client X") — OR use your organization's enterprise AI platform, which typically has contractual data protections in place.

Memory hook: "HIPAA, FERPA, PCI, privilege, proprietary — never paste; always anonymize or use enterprise tools."

Segment 3 — Terms of Service, Data Retention & Business Controls (20 min)

Plain language first. Every AI tool has a terms of service and a privacy policy. Most people do not read them. This segment gives you the three questions to look for so you do not have to read the whole document.

The three questions to check in any AI tool's ToS/Privacy Policy:
1. Does the tool store my inputs? (Almost always yes, at least temporarily — the question is for how long and for what purpose.)
2. Can my inputs be used to train or improve the model? (On many free tiers, yes — unless you opt out. Look for a "Data Controls" or "Privacy" section in your account settings.)
3. Can I opt out — and how? (On most major consumer tools, yes — but you have to find and enable it. It is not usually the default.)

Consumer vs. enterprise/business tiers: the same company may offer very different data protections depending on the tier. Enterprise and business agreements typically include contractual data-processing protections, often including a prohibition on using customer data for training. If your workplace or institution has an enterprise agreement with an AI provider, use that — not your personal free account — for work with sensitive but non-prohibited information.

Anonymizing inputs (data minimization): even when data is not legally protected, it is good practice to share as little identifying information as necessary. Replace names with "Person A," remove specific addresses and dates, use "our company" instead of naming your employer. The AI can still help you effectively with much less information than you might think you need to provide.

Live demonstration: open an approved assistant and show where to find the data-controls or privacy settings in the account. (This will vary by tool — do this step live and narrate what you find, rather than scripting it to a specific menu path that may change.)

Segment 4 — Content Ownership / IP / Copyright + Bias (22 min) · Session 1 closes (~75)

Part A — Content ownership and copyright (15 min)

Important caveat (state this first, explicitly): What follows is an educational overview of a contested and rapidly evolving legal landscape. It is not legal advice. For any specific commercial use, consult a copyright attorney or your institution's counsel — not an AI chatbot.

The core question: if an AI generates content in response to your prompt, who owns it?

The current U.S. landscape (as of 2025):
- The U.S. Copyright Office has repeatedly stated that copyright requires human authorship. Works generated entirely by a machine, without meaningful human creative contribution, are generally not eligible for copyright protection.
- Courts have generally supported this position in early cases involving purely AI-generated art and writing (though this area is actively litigated and will continue to develop).
- The more meaningful human creative input — selecting, arranging, significantly modifying — the stronger the argument for copyright protection in the human's contribution.
- Note: the AI tool's own terms of service may address this. Some tools assert a license to outputs generated using their platform; others disclaim ownership. Read the ToS.

The training-data question (a separate contested issue): whether AI tools trained on copyrighted works without license constitutes infringement is actively litigated. Several major cases are working through the courts. This course does not take a position on the outcome — it is genuinely unresolved.

Practical guidance for students: if you intend to publish, sell, or submit AI-assisted work commercially:
- Do not rely on an AI chatbot's summary of copyright law (ironic but true — verify against official sources: copyright.gov; U.S. Copyright Office guidance).
- Consider your level of human creative contribution and transformation.
- For significant commercial uses, consult an attorney — not an AI.

Part B — Bias and fairness (7 min)

Plain language: AI models are not neutral. They are trained on large datasets that reflect the biases present in human-generated text — which reflects historical inequities, cultural assumptions, and representation gaps. This can result in AI outputs that underrepresent certain groups, reinforce stereotypes, perform differently across languages and demographics, or embed problematic assumptions.

Bias in AI is a subject of active research. The major AI providers have invested significantly in bias mitigation — but none has solved it. Your responsibility as a critical AI user: check outputs for fairness, especially when generating content about people or making decisions that affect people's lives. Verify against diverse sources. Do not assume AI output is representative or equitable by default.

Segment 5 — Academic & Professional Integrity (18 min) · Session 2 opens

Hook back in: "Last session: the privacy and IP rules — the legal and data frameworks. This session: the integrity and ethics rules — how you conduct yourself and build an ethical framework you can actually use."

Academic integrity with AI:

The rule: submitting AI-generated work as your own, when the assignment requires your own thinking and writing, is academic dishonesty. It is the same act as using a ghostwriter without disclosure, or submitting someone else's paper.

The nuance: "using AI" is a spectrum, not a binary. Using AI to brainstorm ideas, check grammar, get feedback, or research a topic — then writing in your own voice and thinking — is different from pasting a prompt and submitting the output verbatim. The line that matters is: whose thinking is being assessed? If the assignment is assessing your analysis, argumentation, and expression, the thinking must be yours.

Every institution has its own AI policy, and they vary. This course's policy is explicit and unusual: AI is required on coursework and banned on quizzes and exams. Know your institution's policy for each course and context.

Professional integrity with AI:

The professional standard is converging on transparency: disclose AI use according to your organization's policy, the platform's requirements, or the publication's guidelines. Many publishers, journals, employers, and clients now require disclosure of AI assistance in submitted work.

The rule behind the rules: you are the author and you are responsible for the accuracy and quality of the final work. AI is the tool; you are the professional who used it. If the AI generated a fact that turned out to be wrong and you published it, that is on you — not the AI.

Memory hook: "The AI drafts; you are the author. Disclose, verify, and take responsibility."

Segment 6 — Troubleshooting: Skill 13 (18 min)

Skill 13 — the four moves:

When AI is not working well, before you give up or assume the technology is broken, run through these four moves:

Move 1 — Start over. A fresh conversation resets the context window. Long conversations accumulate noise and can cause the model to lose track of early instructions, contradict itself, or degrade in quality. Starting fresh is almost always the right first move. Do not be attached to the current conversation — if it is going badly, start over.

Move 2 — Manage memory and context. If you need a long project to stay coherent across multiple sessions, use a tool with project memory (such as Claude Cowork) — or explicitly re-state your key instructions at the start of each new conversation. Think of context management as a skill: front-load your most important instructions and constraints.

Move 3 — Try a different model. Different AI models have genuinely different strengths. If ChatGPT is struggling with a complex coding problem, try Claude or Copilot. If Claude is giving flat creative output, try Gemini. If one model seems stuck in an unhelpful pattern, a different model's different training may approach the task differently. This is not a failure — it is good tool management.

Move 4 — Use AI to teach AI. Paste the failing output back to the same or a different AI and ask: "Here is what I asked for and here is what I got. What went wrong with the response? How should I revise my prompt or approach?" This applies the verification habit you have built all term directly to the AI's own process. It often works remarkably well.

Live demonstration: take a deliberately weak or broken AI output (have one ready), paste it to an approved assistant with the "use AI to teach AI" prompt, and narrate what happens.

Segment 7 — The Future of Work and AI: Competing Views (18 min)

Setting the frame: this segment presents competing views on a genuinely contested question. The goal is not to give you "the answer" — the goal is to give you the thinking tools to form your own considered view and to adapt effectively regardless of how the scenario plays out.

The concern:
AI can now perform tasks that were previously exclusively human — writing, legal research, code review, financial analysis, medical image reading, customer service, and more. Automation will displace some jobs and significantly transform others. Workers whose tasks are most automatable face real risk of displacement, and the transition will not be painless or equitable.

The optimistic counter:
Every major technology transition in history — mechanization, electrification, the internet — initially raised similar concerns about mass unemployment, and in most cases created more jobs than it eliminated (though sometimes in different sectors, over longer timelines, and not without real disruption to particular workers and industries). AI may augment workers more than replace them, especially in tasks requiring judgment, ethics, relationships, creativity, and physical dexterity in complex environments.

The honest synthesis (without a verdict):
Both things can be true simultaneously. Some tasks and some roles will be displaced; some will be transformed; new roles will emerge. The distribution of impact will likely be uneven — by industry, by skill level, by geography, by demographic. The workers best positioned are those with AI fluency (knowing how to use and direct these tools effectively) combined with distinctly human capabilities that are hard to automate.

What you can actually do — adapting to the AI moment:
- Build and maintain AI fluency alongside your domain expertise. The hybrid skill is the durable one.
- Cultivate distinctly human capabilities: judgment, ethics, complex communication, creativity, leadership, physical craft.
- Stay informed: the landscape changes quickly; learning the principles is more durable than memorizing current tool specifics.
- Engage the ethics: be part of the conversation about how AI should be used in your industry and field.

Segment 8 — Building Your Ethical Framework + AI-Critique + Callback & Hand-off (22 min) · Session 2 closes (~75)

The ethical framework — a personal AI Code of Conduct:

A framework is not a list of platitudes. It is a set of specific, tested, operable rules. Your Code of Conduct should answer at minimum:
1. Privacy rule: What will I never paste into a free AI tool? (Name specific categories.)
2. Anonymization rule: When I need to use AI on something sensitive, how will I protect it first?
3. Verification rule: What will I always verify before trusting or sharing AI output?
4. Disclosure rule: When and how will I disclose that I used AI?
5. IP/Copyright rule: How will I handle AI-generated content I want to use or publish?
6. Enterprise rule: When does my work require an enterprise AI tool rather than a consumer one?

The Studio this week gives you the structure to build this document and test it against real scenarios.

AI-critique moment — "AI is not a lawyer" (the course through-line, applied to ethics):

Ask an approved assistant: "What are the legal rules for using AI-generated content commercially? Can I copyright an AI-generated image I sell?"

You will get a confident, well-organized answer. Now apply the verification discipline from Week 10:
- Is this answer current? (Laws change; the AI's training cutoff may predate relevant rulings.)
- Is it accurate? (Cross-check against copyright.gov or the U.S. Copyright Office's published guidance.)
- Does it account for your jurisdiction? (Copyright law varies by country.)
- Did the AI disclose uncertainty? (It should have; if it did not, that is itself a warning sign.)

The lesson: AI can give you a plausible-sounding legal summary that is outdated, incomplete, or wrong. Catch the over-confidence; flag the uncertainty; know where to actually verify.

Callback:
- Everything this term has prepared you for this week: the general→specific mindset (Week 1), verification habits (Week 10), Cowork for enterprise workflows (Weeks 11–14), and now the framework for using all of it responsibly.

Tease next week:
- "Next week: the final exam — cumulative, all objectives — and your capstone. The capstone is your chance to build, document, and verify a real AI-powered workflow that solves a genuine problem. Your Code of Conduct from this week's Studio is the ethical reflection component of your capstone. Start thinking now."

Hand-off (the week's graded work):
- Lecture Tutorial 15 — AI tutor, share-link submission.
- Quiz 15 (no AI), Discussion 15 ("Who Owns AI-Generated Work?"), and Assignment 15 ("Your Responsible AI Framework").
- AI Build Studio 15 — "Your AI Code of Conduct" — build a framework, test it on scenarios, and catch where AI gives over-confident legal/privacy claims.

Instructor FAQ — Common Stumbles

Student says / does	Quick cure
"Pasting client data into a free AI tool is fine if I anonymize afterward."	The order matters: anonymize FIRST, then paste. Never paste identified data and hope to recover it.
"AI said it's copyright-free so I can sell it."	AI is not a legal authority on its own output's copyright status. Verify against official sources; consult an attorney for commercial use. AI is not a lawyer.
"The AI assured me my inputs are private."	AI tools can describe their policies, but they can also get details wrong. Go to the actual ToS/privacy policy on the company's official site.
"Submitting AI text is fine as long as it's accurate."	Academic integrity is about whose thinking is assessed, not just accuracy. If the assignment requires your analysis, the analysis must be yours.
"AI is neutral — it was trained on data."	The training data reflects human-produced text, which reflects existing biases. More data ≠ more neutral.
"My job is safe — AI can't do creative/relationship work."	Present the competing view fairly: some argue uniquely human skills are durable; others argue AI will advance further than expected. Both views deserve to be heard.
Asks the AI for legal advice about HIPAA compliance.	AI is not a lawyer and not a HIPAA compliance officer. For institutional compliance questions, consult your organization's legal counsel or privacy officer.

Scope flag

This outline covers Objective 7 — privacy, ToS, IP, bias, integrity, troubleshooting, and the future of work — at the practical and conceptual level. Legal questions (copyright, HIPAA) are taught factually and with the "not legal advice" caveat; no legal advice is given or implied. Competing views on AI and jobs are presented evenhandedly without a verdict. Real legal frameworks (HIPAA, FERPA, PCI) are named factually; the "not legal advice" caveat is stated explicitly in the lecture. No invented cases, statistics, or citations.

~ Prof. Quinn's edition · Fall 2026 · built with thecoursemaker.com