# Master Teaching Guide How to run any session of this workshop — including with a substitute instructor who has only basic AI knowledge. --- ## If You Are a Substitute: Read This First (15 minutes) You can run this class. Here's the system: 1. **Read this file** (~10 min), then the week's `instructor-guide.md` (~20 min). The weekly guide contains a timed agenda, a full concept walkthrough in plain language, a demo script that tells you what to click and type, and answers to the questions students will actually ask. 2. **You do not need to be an AI expert.** The weekly guide scripts every concept. Where students know more than you about a tool — say so, and put them to work demonstrating. This models exactly the honesty about AI limits that the course teaches. 3. **The demo prompts are pre-written** in `DEMOS_AND_VISUAL_AIDS.md`. You paste them, you don't improvise them. If a tool fails, every weekly guide has a "If the Tool Fails" backup that needs no AI at all. 4. **Your real job is the rhythm:** keep the four blocks moving (concepts → demo → lab → share-out), protect the breaks, and ask the critique questions out loud. Students do the rest. 5. **Never claim AI output is good art.** When a demo result appears, your default line is: "Okay — what did it get wrong?" That question is the whole pedagogy. 6. **The integrity recap script** (below) opens every session. Read it verbatim if you like. **Minimum substitute prep checklist:** - [ ] Read this file + the week's instructor guide - [ ] Open the week's slide deck (`site/weeks/week-XX-slides.html` in any browser; arrow keys to advance) - [ ] Log into one chat AI account (ChatGPT or Claude) and test one demo prompt - [ ] Print or open the week's student handout for your own reference - [ ] Skim "Expected Confusion Points" and "Skeptical Questions" in the weekly guide --- ## Teaching Voice Lead with respect for student skepticism. Do not sell AI as magic. Frame it as a messy, powerful, ethically complicated **production technology** that students deserve to understand. **Do:** - Speak to students as artists and future professionals, not consumers. - Use real creative examples: studios, pipelines, client work, portfolios. - Show failures on purpose. A demo that goes wrong is a better lesson than one that goes right. - Separate, constantly and explicitly: AI assistance vs. authorship, taste, ethics, craft, and disclosure. - Say "I don't know" and "verify that" out loud. Model the behaviour. **Don't:** - Don't use hype language ("revolutionary," "game-changing," "the future is here"). - Don't say "just prompt better" — diagnose *what's missing from the brief* instead. - Don't mock students who refuse to generate images. Their position is defensible and the course defends it. - Don't let enthusiasts present raw output as finished work without being asked the critique questions. - Don't pretend tools are stable. Names, features, and prices change monthly; the decision patterns don't. --- ## First-Day Framing Script (also the weekly recap, condensed) > You do not have to like AI to learn it. You do not have to let it make your art. But if you are entering creative industries after 2026, you need to understand how it changes research, production, jobs, and client expectations. > > In this room: AI is allowed for friction — the boring, repetitive, structural parts. Your taste, your concept, your ethics, and your final craft stay human. Raw AI output is never finished work. We disclose AI use when it matters. And skepticism is welcome whenever it's specific. **5-minute weekly recap (every session, weeks 2–15):** 1. This workshop = leverage without surrendering authorship. 2. AI for friction; humans for meaning. 3. Raw output is raw material, not finished work. 4. Verify before you trust; disclose where it matters. 5. Today's promise: [read the week's one-liner from the slide deck]. --- ## Standard 4-Hour Structure Every weekly instructor guide uses this skeleton. Times are targets, not handcuffs — but protect the lab and the breaks. | Clock | Block | Minutes | What happens | |---|---|---|---| | 0:00–0:10 | Arrive + recap | 10 | Settle, integrity recap, today's promise, show the deliverable | | 0:10–0:50 | Concepts | 40 | Core lesson from slides; one mini-exercise to keep hands moving | | 0:50–1:00 | **Break** | 10 | Non-negotiable | | 1:00–1:45 | Demo | 45 | Live workflow with pre-written prompts; students follow or watch; ends with a 5-min "your turn" | | 1:45–1:55 | **Break** | 10 | Non-negotiable | | 1:55–3:05 | Lab | 70 | Constrained activity, per-track variants; instructor circulates | | 3:05–3:15 | **Break** | 10 | Non-negotiable | | 3:15–3:50 | Share + critique | 35 | 3–6 volunteers; critique questions; peers respond | | 3:50–4:00 | Reflect + wrap | 10 | Reflection prompts, deliverable check, next week teaser | **Why breaks are sacred:** 4 hours is long, the audience is young, many are neurodivergent, and lab energy collapses without resets. If you're running over, cut concept content — never the break, never the lab. **Timing flexibility rules:** - Demo running long → cut the optional demo segment (each weekly guide marks one). - Lab running hot → let it run 10 extra minutes, trim share-out to 25. - Low attendance (<6) → convert share-out into a roundtable; everyone shares small. - High attendance (>20) → share-out in pairs first, then 3 volunteers to the room. --- ## Facilitation Playbook ### Running the concept block - Slides carry the structure; the weekly guide carries your words. Don't read slides aloud — say the talking point, let them read the slide. - Every 10–12 minutes, drop in the marked "Try this now" micro-prompt (30–90 seconds, phones allowed). It resets attention. - Definitions appear on slides when first used. Point at them. The handout repeats them — say so, so nobody panics about note-taking. ### Running the demo - Pre-paste prompts into a doc you can copy from (all in `DEMOS_AND_VISUAL_AIDS.md`). - Narrate the *decisions*, not the typing: "I'm giving it constraints because last time it went generic." - When output appears: pause. Ask the room: **"What did it get wrong?"** Collect three answers before you comment. - Show one deliberate failure per demo (each weekly guide includes one). The lesson is the recovery. - Keep a screenshots folder of past outputs as backup if the live tool is down. ### Running the lab - Write the lab constraint on the board/slide and leave it up the whole time. - Circulate in a loop. Ask, in order: "What are you making?" → "What did AI do?" → "What are you deciding?" - Students stuck staring at outputs → force a rejection round: "Kill four of the six. Tell me why." - Students racing ahead → point them to the handout's Extension activity. - Students refusing generation → point them to the zero-generation track (every lab has one). - 10-minute and 2-minute warnings, out loud. ### Running critique Use the same five questions every week (students learn to pre-answer them): 1. Where is the human point of view in this? 2. What did the model make generic, and did you catch it? 3. What did you reject, and why? 4. What would make this portfolio-safe (craft + disclosure)? 5. What would you verify before showing a client? Critique tone: we critique *the workflow and the choices*, not the student, and not the model's aesthetics — the model doesn't care. --- ## Classroom Norms (post these) 1. Raw AI output is not finished work. 2. Process documentation matters — screenshots, prompts, decisions. 3. Tool failure is part of the lesson, not an interruption of it. 4. Skepticism is welcome when it is specific. 5. Nobody is required to generate images — planning/critique tracks always exist. 6. We don't clone living artists, classmates, or anyone's voice/face. 7. What you put into a tool may leave the room — no private, client, or classmate material without permission. 8. Phones/laptops are tools here, not contraband. Use them on the work. --- ## Handling the Room ### Skeptical or hostile students - First move: agree with the legitimate part. ("You're right that training data consent is a real, unresolved problem.") - Redirect to specificity: "Which use bothers you — research summaries, image generation, or style mimicry? They're different cases." - Offer the refusal-with-knowledge framing: understanding the machine ≠ endorsing it. Critics who understand the technology are *more* credible, not less. - Never spend more than 3 minutes of plenary time in a debate; offer to continue at break and put the question in the share-out. ### Anxious students ("will I have a job?") - Don't dismiss it and don't catastrophize. Honest line: "Some tasks are getting automated. The people thriving are the ones whose judgment, taste, and accountability sit on top of the tools. That's learnable, and it's what this room practices." - Point to what AI reliably can't do this decade: be accountable, hold a client relationship, have taste developed from lived experience, sign their name to the result. ### Overwhelmed students - One instruction at a time, in writing (the handout has the steps — point to the exact block). - Pair them with a patient peer, not the fastest one. - Remind: the deliverable is small and achievable in the time given. The handout works as a self-paced fallback. ### The student who knows more than you - Deputize them. "Show us." Then ask them the integrity questions about their own workflow — that's usually new to them. --- ## Accessibility and Accommodation Notes - **Materials are ADHD-friendly by design:** chunked blocks, strong headings, checklists, repeated structure across handout/site/slides. Don't undo this by improvising long unstructured lectures. - **All instructions appear in writing** (slides + handout), never voice-only. Repeat key instructions verbally twice. - **Screen readers / low vision:** site pages are semantic HTML; encourage zoom; slide decks readable as single scrolling pages (print mode). - **Captions:** any video shown in class should have captions on. Embedded YouTube videos: turn CC on before class. - **Processing time:** after asking a question, count five seconds of silence before rescuing it. - **Breaks are accommodation infrastructure** — students who need to move/leave can sync with them; nobody needs to ask permission to step out. - **Free-tier students:** never single out who is on which plan; labs are pair-friendly by default. - **Opt-outs:** image-generation opt-out is permanent and penalty-free; the zero-generation track produces the same deliverable category every week. --- ## Drop-In Adaptation Each week must stand alone: - Begin with the 5-minute integrity recap (script above). - Never assume artifacts from previous weeks; labs include a "cold start" entry point (weekly guides flag it). - End with a concrete artifact students keep even if they never return. - Returning students stack artifacts toward the Week 14–15 capstone; the weekly guides note how. --- ## When Tools Fail (general protocol) 1. Stay calm and narrate: "This is the lesson — production AI fails, professionals have fallbacks." 2. Switch to the backup in the weekly guide (screenshots of prior runs, prepared outputs, or the paper version). 3. Lab continues in critique/planning mode: students can always analyze prepared outputs, write briefs, build checklists, or storyboard on paper. 4. If wifi is down entirely: every weekly guide has a full offline variant of the lab. 5. Log what failed in `PRODUCTION_NOTES.md` so the next run pre-checks it. --- ## What Administrators Should See (any week) If the Dean drops in unannounced, any session should visibly demonstrate: - A stated learning objective on screen and a timed structure being followed. - Students producing work and **rejecting AI output with reasons** — taste being trained, not bypassed. - Integrity language in active use (disclosure, verification, consent). - Differentiation: per-track lab variants and a zero-generation option. - An artifact every student leaves with. Each weekly instructor guide ends with an "Admin Observation Notes" block naming what that specific week demonstrates pedagogically. The full rationale lives in `ADMIN_OBSERVATION_BRIEF.md`. --- ## Suggested Final Capstone A personal AI-enhanced creative workflow, tiny tool, prototype, process board, or portfolio/career system that saves time or improves output **without replacing the student's creative voice**. Scoped in Week 11–13, built in Week 14, presented in Week 15. Capstone quality bar (shared with students in Week 14): 1. Solves a real friction in *their* practice. 2. Shows documented human decisions (selection, rejection, refinement). 3. Includes honest disclosure of what AI did. 4. Would survive the student demoing it to a future employer.