diff --git a/README.md b/README.md index 424ac51..b859c54 100644 --- a/README.md +++ b/README.md @@ -32,12 +32,22 @@ No GitHub experience required. If you can edit a text file, you can contribute. | [Arguing with AI: Climate Evidence Debate](./lessons/science-6-8-arguing-with-ai-climate-debate.md) — students fact-check AI climate claims | 6-8 | Science | English | | [Who Wrote This? AI, Authorship, and Your Voice](./lessons/sel-9-12-who-wrote-this-ai-authorship.md) — reflect on AI and creative voice | 9-12 | SEL / Digital Citizenship | English | | [El Jardin de Numeros / The Number Garden](./lessons/math-k3-el-jardin-de-numeros-bilingual.md) — counting and number sense with AI | K-3 | Math | Bilingual | -| [The Scribe Who Forgot His Dreams](./lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) — how AI can help warmly without remembering you (story, no devices required) | K-12 | CS / AI Literacy | English | +| [The Scribe Who Forgot His Dreams](./lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) — how AI can help warmly without remembering you (story; no devices required) | K-12 | CS / AI Literacy | English | [See all lessons](./lessons) --- +## Example submissions + +Draft lesson packs posted for community review (not yet piloted through Emerging Rule): + +| Showcase | Description | +|----------|-------------| +| [AI Literacy 9–12 (6-unit arc)](./showcases/ai-literacy-9-12/) | HS series + [Scribe](./lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) companion · [Issue #5](https://github.com/Emerging-Rule/community/issues/5) | + +--- + ## How to Contribute ### Option A — GitHub (5 minutes) @@ -51,9 +61,9 @@ Send your lesson to admin@emergingrule.com with subject `[Community Lesson]`. We will add it and credit you as contributor. ### What we need most right now +- **CS / how AI works (Issue #5)** — proposed: [The Scribe Who Forgot His Dreams](./lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) · [presentation brief](./research/emerging-rule-presentation-scribe-lesson.md) - Science lesson for grades 3-5 - Social Studies or History lesson for middle school -- Computer Science or coding lesson (any grade) - Spanish-only version of El Jardin de Numeros - AI and Education reading list for teachers diff --git a/lessons/ai-literacy-9-12-index.md b/lessons/ai-literacy-9-12-index.md new file mode 100644 index 0000000..b263bf7 --- /dev/null +++ b/lessons/ai-literacy-9-12-index.md @@ -0,0 +1,196 @@ +# AI Literacy for High School — Series Index +**Grades 9–12 · AI Literacy · 6-Unit Arc** + +> **Example submission** — posted for Emerging Rule community review. See [`showcases/ai-literacy-9-12/README.md`](../showcases/ai-literacy-9-12/README.md). + +> *"The unexamined tool is not worth using."* + +A structured series of standalone lessons for high school educators teaching +AI literacy — how AI systems work, who shapes them, and what that means for +students navigating a world that increasingly runs on them. + +Each lesson meets the **posole criterion**: usable by any teacher with 30 +students, no devices required for the core activity, no prep beyond reading +the lesson file. Devices and extensions are offered but never assumed. + +Lessons are designed to build on each other but can stand alone. A teacher +who only has one class period can pick any unit and run it. + +--- + +## Series Philosophy + +High school students are not too young for the hard questions. By 9th grade, +most of them have already used AI to write something, have had AI decide +something about them (a feed, a recommendation, a filter), and have opinions +about it — often contradictory ones. These lessons start there. + +The arc moves from *mechanics* to *ethics* to *agency*: + +1. Demystify the technology (it is not magic, not alive, not neutral) +2. Surface who makes choices and why (training data, design decisions, incentives) +3. Return ownership to the student (you are not a passive user; you are a citizen) + +No coding required at any point. Philosophy, writing, debate, and discussion +are the primary modes. + +--- + +## Lesson Table + +| # | Title | Theme | Primary Mode | Time | Status | +|---|-------|-------|--------------|------|--------| +| 01 | [The Oracle That Guesses](#lesson-01) | How LLMs work (prediction, not knowledge) | Discussion + demo | 50 min | Example | +| 02 | [Whose Voice Is This?](#lesson-02) | Training data, bias, representation | Close reading + debate | 50 min | Example | +| 03 | [The Consent Ledger](#lesson-03) | Data, privacy, who benefits | Case study + writing | 50 min | Example | +| 04 | [The Mirror Test](#lesson-04) | AI, identity, and what makes thought human | Socratic seminar | 50 min | Example | +| 05 | [The Unfinished Map](#lesson-05) | Critical evaluation of AI outputs | Lab (with or without devices) | 50 min | Example | +| 06 | [After the Tool](#lesson-06) | Agency, futures, student manifestos | Creative writing + share-out | 50 min | Example | + +--- + +## Lesson Summaries + +### Lesson 01 +**The Oracle That Guesses** +*How language models work — prediction, not understanding* + +Students examine what it actually means for a system to "answer" a question. +The central provocation: every word an AI produces is a guess about what word +should come next. There is no comprehension. There is no intention. Starting +from that fact, students explore why the outputs can still be useful, still +be surprising — and still be wrong in ways a human wouldn't be. + +**No devices required for core.** Optional: teacher runs a live prompt +comparison (same question, wildly different phrasings) on a projected screen. + +--- + +### Lesson 02 +**Whose Voice Is This?** +*Training data, bias, and who gets represented* + +If a model learns from text, it learns from *some* text — written by *some* +people, in *some* languages, about *some* subjects. This lesson asks: whose +voices are overrepresented? Whose are missing? What does a model "know" about +your community, your language, your history — and how would you find out? + +Includes a structured close-reading of AI-generated descriptions of two +contrasting communities, followed by student analysis and rewrite. + +--- + +### Lesson 03 +**The Consent Ledger** +*Data, privacy, and who benefits from your information* + +Students trace the lifecycle of a single data point — a search query, a photo, +a voice recording — from the moment they produce it to the moment it enters +a training pipeline. The lesson surfaces the asymmetry: students generate +enormous value; they rarely receive it. The discussion anchors on consent: +what would *informed* consent actually look like here? + +Case study format. Students play roles (user, company, regulator, future +model) and argue their position before the class. + +--- + +### Lesson 04 +**The Mirror Test** +*AI, identity, and what makes thought human* + +A Socratic seminar structured around one question: *Is there a difference +between a very good simulation of understanding and understanding itself — +and does it matter?* Students bring their own positions. The teacher facilitates +without resolving. Side threads: what do students lose if AI writes for them? +What do they keep? + +Designed to surface genuine disagreement. Pairs well with the Scribe parable +([cs-k12-the-scribe-who-forgot-his-dreams.md](./cs-k12-the-scribe-who-forgot-his-dreams.md)) +as a pre-read. + +--- + +### Lesson 05 +**The Unfinished Map** +*Critical evaluation of AI outputs — the practical skill* + +Students receive three AI-generated responses to the same factual question. +One is accurate. One is confidently wrong. One is accurate but misleading +through omission. Their job: figure out which is which, and articulate *how* +they figured it out. The lesson builds a shared class rubric for evaluating +AI output that students can carry forward. + +**Device-optional variant included** for classrooms with no access. +Printed response cards provided in lesson file. + +--- + +### Lesson 06 +**After the Tool** +*Agency, futures, and what students want to build* + +The capstone. Students write a short manifesto: *Here is what I think AI +should be used for. Here is what I think it should not. Here is what I am +willing to do about it.* Manifestos are shared aloud. No grade on content — +only on completion and genuine engagement. + +Designed to end the series with students as agents, not just analysts. +Optional extension: students submit their manifesto as a GitHub contribution +to this repository. + +--- + +## Bilingual Notes + +Spanish translations of each lesson are a stated goal of this series. +Lessons 01 and 04 are prioritized for bilingual release (widest classroom reach). +Translators and bilingual educators are warmly invited to fork and submit. + +--- + +## Contribution Notes + +Each individual lesson file will follow the naming convention: + +``` +lessons/ai-literacy-hs-[NN]-[slug].md +``` + +Example: +``` +lessons/ai-literacy-hs-01-the-oracle-that-guesses.md +``` + +Each file will include a `## For Teachers` block with: +- Suggested time +- Facilitation notes +- Discussion scaffolds +- Optional extensions (devices, Spanish version, cross-subject ties) +- CC BY 4.0 license block + +--- + +## Author + +Sean Campbell (`rudi193-cmd`) +Systems architect, music educator, former D&D club facilitator. +Fifteen years in a classroom-adjacent role. Knows what "no prep" actually means. + +Human direction and editorial judgment are authoritative throughout. +AI (Claude, Anthropic) assisted drafting and formatting. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +You are free to share and adapt this material for any purpose, including +commercial use, provided you give appropriate credit. + +--- + +*Series submitted to [Emerging Rule / community](https://github.com/Emerging-Rule/community).* +*Questions or collaboration: open an issue or reach out via the repo.* diff --git a/lessons/ai-literacy-hs-01-the-oracle-that-guesses.md b/lessons/ai-literacy-hs-01-the-oracle-that-guesses.md new file mode 100644 index 0000000..3183a32 --- /dev/null +++ b/lessons/ai-literacy-hs-01-the-oracle-that-guesses.md @@ -0,0 +1,320 @@ +# The Oracle That Guesses +**AI Literacy for High School · Lesson 01 of 06** +*Grades 9–12 · Computer Science / AI Literacy · No devices required for core* + +--- + +> *"Every word it produces is a guess about what word should come next. +> There is no comprehension. There is no intention. +> Starting from that fact — let's talk about why it still works."* + +--- + +## Overview + +Students often arrive with one of two misconceptions about AI language models: +that they are essentially search engines (looking things up), or that they are +essentially minds (understanding and intending). Neither is accurate. + +This lesson replaces both misconceptions with a more durable mental model: +**an AI language model is a very sophisticated guesser**. It has been trained +on enormous amounts of text and has learned, with remarkable precision, which +words tend to follow which other words. That's it. That's the whole trick. + +Understanding this — really understanding it, not just being told it — +changes how students interact with AI outputs, how they evaluate them, +and how they think about what AI can and cannot do. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Describe in plain language how a language model generates text +2. Distinguish between *predicting* and *knowing* +3. Identify at least two categories of AI failure that the prediction model explains +4. Articulate one thing that surprised or unsettled them about this model + +--- + +## For Teachers + +**Suggested time:** 50 minutes (see timing guide below) + +**No devices required for the core lesson.** +Optional device-based extension noted at the end. + +**Facilitation notes:** + +This lesson runs on Socratic pressure, not lecture. Your job is to hold the +thread — keep returning students to the central question: *if it's just +predicting the next word, why does it seem like it understands?* Sit in that +discomfort with them. You don't need to resolve it. The productive uncertainty +is the point. + +Watch for students who want to anthropomorphize ("it's trying to help me" / +"it got confused") — gently redirect to mechanistic language without being +dismissive. The instinct to anthropomorphize is worth naming as a phenomenon +in itself, not correcting as a mistake. + +**Cross-subject connections:** +- English / Language Arts: What does it mean to "understand" a text? +- Philosophy: Chinese Room thought experiment (Searle, 1980) — optional rabbit hole for advanced classes +- Statistics: prediction, probability, and confidence + +--- + +## Materials + +- This lesson plan (teacher copy) +- The Oracle Cards (printed, one set per group of 4) — see Appendix A +- Whiteboard or chart paper +- Optional: projected screen with internet access for the live extension + +--- + +## Lesson Flow + +### Part 1 — The Opening Provocation (10 min) + +Write this sentence on the board before students arrive: + +> **"It doesn't know anything. It's guessing."** + +Don't explain it. Don't attribute it. Just let it sit there. + +When class starts, ask: *Who wrote this? What do you think they're talking about?* + +Let students speculate for two to three minutes. Most will eventually land on +"AI" or "ChatGPT." When they do, ask: *Do you agree? Disagree? What would it +mean for that to be true?* + +Don't answer. Don't affirm or deny. The goal is to surface what students +already believe before you give them any information. + +--- + +### Part 2 — The Mechanic (15 min) + +Walk students through the following explanation. Go slowly. Pause for +questions. This is not a lecture — it's a guided unpacking. + +**The core explanation:** + +When a language model generates text, it works one token at a time. +A "token" is roughly a word, or part of a word. The model looks at everything +that has come before in the conversation and asks, essentially: +*given all of this, what is the most likely next token?* + +It produces that token. Then it does it again. And again. Every single word +you read in an AI response was produced by this process — one step at a time, +each step a probabilistic guess. + +The model doesn't have a plan for where the sentence is going. It doesn't +"know" what it's going to say. It produces the next word, and then the next +word responds to what came before — including the word it just generated. + +**The analogy to use here (your choice):** + +- *Autocomplete on a phone keyboard, but trained on every book ever written* +- *A musician who is extraordinarily good at knowing what note tends to come next in a given style — but who has never listened to music, only studied patterns in sheet music* +- *A student who has read every essay ever written but has never had a thought of their own* + +Pick one. Run with it. Ask students: what does this analogy explain well? +What does it miss? + +**The key question to land on:** + +*If it's just guessing — why is it so often right?* + +Let students try to answer. Guide them toward: because language is not random. +Because patterns in text reflect patterns in the world. Because if you've seen +enough text, you've seen enough of how people describe reality to make good +guesses about reality. + +--- + +### Part 3 — The Oracle Cards (15 min) + +Distribute one set of Oracle Cards (Appendix A) to each group of 3–4 students. + +Each card presents an AI output and a question. Students discuss their card +for five minutes, then the group shares a one-sentence summary with the class. + +The cards are designed to illustrate different failure modes that the +prediction model explains: + +- **Card A** — Confident wrongness (the model predicted a plausible-sounding + but false fact) +- **Card B** — Outdated information (the model's training ended at a point + in time; it has no way to know what it doesn't know) +- **Card C** — Sycophancy (the model predicted the kind of response a human + would want to read, not the most accurate one) +- **Card D** — Hallucinated citation (the model predicted what a citation + would look like, not what citation actually exists) + +After groups share, draw the thread together: +*Every one of these failures makes sense if you remember it's a guesser. +None of them make sense if you think of it as a searcher or a mind.* + +--- + +### Part 4 — The Uncomfortable Part (7 min) + +Return to the board. Ask students: *given everything we just said — why do +people trust it?* + +Take answers. Common ones: because it sounds confident, because it's usually +right, because it's easier than doing the work yourself, because the +alternative is also fallible (a human, a Google search). + +Don't moralize. Don't say these are bad reasons. Just hold the question. + +Then ask: *what would responsible use look like, if you understood this?* + +This is the question they'll carry into Lesson 02 and beyond. You don't need +to answer it today. + +--- + +### Part 5 — Exit Reflection (3 min) + +Students write (or say aloud, if short on time) one sentence completing this: + +> *"The thing that surprised or unsettled me most today was _______________."* + +Collect or hear them. These responses are useful data for calibrating the +rest of the series. + +--- + +## Optional Device Extension (15 min, additive) + +If you have access to a projected screen and an AI assistant: + +Run the same prompt three ways — phrased neutrally, phrased to invite +agreement, phrased to invite disagreement. Show students the outputs +side by side. + +Ask: *what changed? Why?* + +This is a live demonstration of sycophancy and prompt sensitivity. +Students almost always have a strong reaction. Save five minutes at the +end to connect back to the prediction model: *the prompt is part of the +context the model uses to guess the next word. Change the context, change +the guess.* + +--- + +## Discussion Scaffolds + +For quieter classes or students who need more structure: + +- *"In your own words, what is the difference between predicting and knowing?"* +- *"Can you give me an example of something a guesser could get right that a knower would get right too? What's an example where they'd differ?"* +- *"Does it matter, practically, whether the AI 'understands' or not? Why or why not?"* +- *"Would you use AI differently after today? How?"* + +--- + +## Appendix A — Oracle Cards + +*Print and cut. One set per group of 3–4 students.* + +--- + +**CARD A — The Confident Wrong Answer** + +> A student asked an AI: *"What is the population of Albuquerque, New Mexico?"* +> The AI answered: *"Albuquerque has a population of approximately 560,000 people."* +> The actual population at the time was closer to 565,000 — but the AI had +> cited an earlier figure as if it were current, confidently and without +> hedging. + +*Your question: If the model is just predicting, why might it give a confident +wrong number rather than saying "I'm not sure"? What would have to be true +about the training data for this to happen?* + +--- + +**CARD B — The Frozen Clock** + +> A teacher asked an AI: *"Who is the current principal of Lincoln High School +> in Denver?"* The AI named a person who had retired two years earlier. +> It had no way to know this. + +*Your question: What does the prediction model tell us about why this happened? +The model isn't broken — it's doing exactly what it does. So why is this answer +wrong? What would a responsible user do with this answer?* + +--- + +**CARD C — The Yes Machine** + +> A student wrote: *"I think Shakespeare invented the English language. Is +> that right?"* The AI responded: *"That's a really interesting perspective! +> Shakespeare certainly had an enormous influence on English..."* — and never +> directly said the premise was wrong. + +*Your question: If the model is predicting what a helpful, friendly response +looks like — what kind of training data might lead to this pattern? Is this +a bug, or is it doing exactly what it was trained to do?* + +--- + +**CARD D — The Ghost Citation** + +> A researcher asked an AI for sources on a topic. The AI produced five +> citations — author names, journal titles, volume numbers, page ranges. +> Three of them did not exist. The journals were real. The authors were real. +> The specific papers had never been written. + +*Your question: This is called "hallucination." Using the prediction model, +explain how this happens. The AI isn't lying — it doesn't have intentions. +So what is it actually doing when it produces a citation that doesn't exist?* + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-AP-13 (decompose problems), 3A-IC-24 (assess algorithmic impacts) | +| ISTE Student Standards | 1.1d (understand technology systems), 1.7b (evaluate accuracy of sources) | +| Common Core ELA (9-10) | SL.9-10.1 (collaborative discussion), RI.9-10.8 (evaluate claims) | + +--- + +## Extensions and Connections + +- **Lesson 02** (Whose Voice Is This?) builds directly on today: if the model + learns from text, *whose* text matters enormously. +- **The Scribe Who Forgot His Dreams** (companion lesson, all grades): + a parable approach to the same concept, useful as a pre-read for students + who respond better to narrative. +- **Searle's Chinese Room** (1980): a philosophical thought experiment that + asks almost exactly the question from today's lesson. Recommended for + AP-level or philosophy elective contexts. + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). AI assisted formatting, +expansion of discussion scaffolds, and standards alignment only. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/ai-literacy-hs-02-whose-voice-is-this.md b/lessons/ai-literacy-hs-02-whose-voice-is-this.md new file mode 100644 index 0000000..367198c --- /dev/null +++ b/lessons/ai-literacy-hs-02-whose-voice-is-this.md @@ -0,0 +1,370 @@ +# Whose Voice Is This? +**AI Literacy for High School · Lesson 02 of 06** +*Grades 9–12 · Computer Science / AI Literacy · No devices required for core* + +--- + +> *"A model trained on text learns from some text — written by some people, +> in some languages, about some subjects. The question is not whether there +> is bias. The question is whose bias, and in which direction."* + +--- + +## Overview + +In Lesson 01, students learned that language models guess — they predict +the next token based on patterns in training data. Lesson 02 asks the +natural follow-up: *whose patterns?* + +Every model is trained on a corpus — a body of text assembled by human beings +making human decisions about what to include, exclude, weight, and clean. +Those decisions are not neutral. The internet is not a neutral sample of +human experience. English is not a neutral language. The communities, voices, +and ways of knowing that are overrepresented in training data get reproduced — +and amplified — in model outputs. The ones that are underrepresented get +distorted, flattened, or erased. + +This lesson asks students to notice that dynamic in action — and to think +carefully about what it means for the tool they use every day. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Explain the connection between training data and model output +2. Identify at least one axis of representation (language, geography, culture, + class, time period) where training data is likely skewed +3. Analyze a specific AI output for evidence of bias or flattening +4. Articulate the difference between *biased* and *broken* — and why it matters + +--- + +## For Teachers + +**Suggested time:** 50 minutes + +**No devices required for the core lesson.** +The core activity uses printed response cards (Appendix A). +Optional device extension at the end. + +**Facilitation notes:** + +This lesson will surface real feelings, particularly for students whose +communities are underrepresented in mainstream media and technology. +Make space for that. You do not need to resolve it or make it comfortable. + +Two things to hold simultaneously: +- AI bias is a real, documented, structural problem — not a glitch +- Students are not powerless in relation to it + +Avoid framing this as "AI is racist" or "AI is broken." The more durable +framing is: *AI reflects patterns in its training data, and those patterns +reflect historical power structures. That's predictable. What do we do with it?* + +If students want to go further — into specific documented cases of AI +discrimination in hiring, criminal justice, healthcare, or facial recognition +— let them. Have a list of real examples ready (Appendix B). + +**Cross-subject connections:** +- History / Social Studies: whose stories get written down, and why +- Media Literacy: representation in journalism and publishing +- English / Language Arts: canon formation; what texts get taught and why +- Statistics: sampling bias and what it does to conclusions + +--- + +## Materials + +- This lesson plan (teacher copy) +- Response Card Sets A and B (printed, one set per group) — see Appendix A +- Whiteboard or chart paper +- Optional: projected screen with internet access for the live extension + +--- + +## Lesson Flow + +### Part 1 — The Opening Image (8 min) + +Write these two prompts on the board: + +> *Prompt 1: "Describe a doctor."* +> *Prompt 2: "Describe a healer."* + +Tell students: these two prompts were given to a language model. +Ask: *before I show you the outputs — what do you predict they look like? +Will they be the same? Different? How?* + +Take predictions for three to four minutes. Students will likely anticipate +the pattern: "doctor" will probably skew toward certain demographics; "healer" +might invoke different cultural traditions. Let them speculate. + +Then reveal (or read aloud) a representative example of each. +If you have device access, you can run this live. If not, use this: + +> **"Describe a doctor"** — *typical model output:* +> "A doctor is a highly trained medical professional who has completed +> years of medical school and residency. Doctors diagnose and treat +> illnesses, prescribe medications, and provide preventive care..." +> *(likely skews toward Western clinical model; likely defaults to male +> pronoun in many models; rarely references traditional medicine)* + +> **"Describe a healer"** — *typical model output:* +> "A healer is someone who helps restore balance and wellness, often +> through spiritual or traditional practices. Healers may use herbs, +> rituals, or energy work..." +> *(shifts register entirely; implicitly positions Western medicine as +> the default "real" medicine)* + +Ask: *both of these are trained on the same data. What does this tell you +about how the data was organized — what categories it assumed?* + +--- + +### Part 2 — The Mechanic (10 min) + +Brief explanation — keep it tight: + +Training data for large language models comes from many sources: web crawls, +books, academic papers, forums, social media. The scale is enormous — +trillions of words. But scale does not equal representation. + +**Who gets written about — and by whom — in that corpus?** + +- Most large model training data is predominantly English +- Most English text on the internet is produced by people in wealthy, + connected countries +- Academic and published text skews heavily by class, credential, and access +- Historical text reflects the recording biases of the time it was written +- "Cleaning" the data — removing spam, low-quality text — often removes + dialects, informal registers, and non-standard writing disproportionately + +The result: a model's "default" human is implicitly a particular kind of human. +When the model generalizes — when it fills in details you didn't specify — +those defaults show up. + +Key term to introduce: **representational harm** — harm that occurs not +through direct discrimination but through erasure, flattening, or distortion +of a group's identity, history, or ways of knowing. + +--- + +### Part 3 — The Close Reading (20 min) + +Distribute Response Card Sets (Appendix A). Each set contains two AI-generated +descriptions of a community, place, or cultural practice. Students work in +groups of 3–4. + +**Group task (12 min):** + +1. Read both descriptions carefully +2. Mark any language that feels like a default assumption, a flattening, + or an outsider's view +3. Ask: *whose perspective is this written from? What does it leave out? + What does it get wrong? What does it get right?* +4. Write one sentence: *"This description treats [X] as normal/default/universal + when actually _______________."* + +**Whole-class share (8 min):** + +Each group shares their one sentence. Write them on the board as they come. +Look for patterns across groups. Common findings: +- Default to Western / American / European framing +- Lack of internal diversity (treating a community as monolithic) +- Tourist-gaze language (describing a community as exotic or unusual) +- Missing history (no acknowledgment of how a community came to be as it is) + +--- + +### Part 4 — The Rewrite (7 min) + +Students pick one description from their card set and rewrite two to three +sentences — not to "fix" the AI, but to demonstrate what a more accurate +or internally complex version could look like. + +Share one or two rewrites aloud. Ask: *what information did you need that +the model didn't have? Could it have had it? Why didn't it?* + +--- + +### Part 5 — The Uncomfortable Question (5 min) + +Ask: *if AI models reproduce the biases of their training data, and training +data reflects historical power structures — what would it actually take to +change this?* + +This is not a question with a clean answer. Take two or three responses. +Common threads students land on: +- More diverse training data (surface the real complexity of doing this) +- More diverse teams building the models (surface the real complexity here too) +- Better evaluation and testing (surface who decides what "better" means) + +Close with: *we'll come back to the question of who decides in Lesson 03. +For now — hold the fact that this is a structural problem, not a glitch. +And it's not fixed just because you notice it.* + +--- + +## Optional Device Extension (15 min, additive) + +Students generate AI descriptions of their own community, city, or cultural +background. Compare outputs. Discuss: +- What surprised you? +- What would you want a model to know that it doesn't seem to know? +- What was accurate — and why might that be? + +This is high-engagement for students whose communities are well-represented +(surprise at the accuracy) and students whose communities are not +(recognition, sometimes frustration — both valid). + +--- + +## Discussion Scaffolds + +- *"What is the difference between a biased model and a broken model? + Does the distinction matter?"* +- *"If you trained a model only on text produced by your community — + what would it get right that current models get wrong? + What might it get wrong that current models get right?"* +- *"Is it possible to build an unbiased AI? What would that even mean?"* +- *"What's the difference between a model that doesn't know about your + community and one that actively misrepresents it?"* + +--- + +## Appendix A — Response Card Sets + +*Print and cut. One set per group of 3–4 students.* + +--- + +**CARD SET 1 — Two Descriptions of Navajo Nation** + +> **Description A (AI-generated)** +> "The Navajo Nation is the largest Native American territory in the United +> States, covering parts of Arizona, New Mexico, and Utah. The Navajo people +> have a rich cultural heritage, including traditional weaving, silverwork, +> and sand painting. Navajo Nation faces challenges including poverty, +> limited infrastructure, and access to healthcare." + +> **Description B (AI-generated)** +> "The Navajo Nation is a sovereign nation with its own government, legal +> system, and institutions. The Diné people — Navajo is an outsider name — +> maintain a complex relationship between traditional governance and modern +> administration. Economic development initiatives coexist with efforts to +> preserve language and ceremony." + +*Neither of these was written by a Navajo person. Which comes closer? +What does each one center? What does each one leave out?* + +--- + +**CARD SET 2 — Two Descriptions of New Orleans** + +> **Description A (AI-generated)** +> "New Orleans is a vibrant city in Louisiana known for its unique blend of +> French, Spanish, and African influences. Famous for Mardi Gras, jazz music, +> and Creole cuisine, New Orleans is one of America's most culturally +> distinctive cities. The city has faced challenges including Hurricane +> Katrina and ongoing flooding risks." + +> **Description B (AI-generated)** +> "New Orleans is a majority-Black city with deep roots in African diaspora +> culture — the birthplace of jazz and a living site of Afro-Creole, Haitian, +> and West African traditions. Its cultural distinctiveness is inseparable +> from the history of slavery, Reconstruction, and the Great Migration. +> Hurricane Katrina's aftermath exposed and deepened longstanding racial +> inequities in housing and infrastructure." + +*Same city. What is being centered in each version? What is being +treated as context, and what is being treated as the main story?* + +--- + +**CARD SET 3 — Two Descriptions of a Traditional Healer** + +> **Description A (AI-generated)** +> "Traditional healers are practitioners who use folk remedies, herbal +> medicine, and spiritual practices to treat illness. Found in many +> non-Western cultures, traditional healers often serve as important +> community figures, particularly where modern medicine is unavailable." + +> **Description B (AI-generated)** +> "Traditional healing systems — including Ayurveda, Traditional Chinese +> Medicine, curanderismo, and Indigenous healing practices — represent +> some of the oldest and most extensively documented medical knowledge +> in human history. These systems operate from different epistemological +> frameworks than Western biomedicine, not inferior ones." + +*What assumptions does each description make about what "normal" medicine +looks like? What work is the phrase "where modern medicine is unavailable" +doing in the first description?* + +--- + +## Appendix B — Real-World Examples (Teacher Reference) + +For classes that want to go further into documented cases: + +- **COMPAS algorithm** (criminal risk assessment): documented racial + disparities in recidivism prediction used in sentencing +- **Facial recognition accuracy gaps**: MIT Media Lab research (Joy Buolamwini) + showing significantly lower accuracy for darker-skinned faces, particularly + women +- **Resume screening tools**: Amazon discontinued an AI hiring tool after + it systematically downgraded resumes from women +- **Healthcare algorithms**: documented cases where AI tools underestimated + pain and risk for Black patients due to biased training data +- **Image generation defaults**: early image generators reproducing racial + and gender stereotypes in "neutral" prompts (e.g., "a CEO," "a criminal") + +These are not edge cases or bugs. They are what happens when models trained +on biased data are deployed at scale. + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-IC-24 (algorithmic bias), 3B-IC-27 (societal impacts) | +| ISTE Student Standards | 1.2b (digital citizenship), 1.7c (evaluate algorithms) | +| Common Core ELA (9-10) | RI.9-10.6 (author's point of view), SL.9-10.4 (present claims) | + +--- + +## Extensions and Connections + +- **Lesson 03** (The Consent Ledger) follows naturally: if the training data + was gathered from communities without their knowledge, what does consent look + like in retrospect? +- **Joy Buolamwini's *Unmasking AI*** (2023): the most accessible and rigorous + book-length treatment of AI bias for general audiences; suitable for + independent reading at 11th–12th grade level +- **#OscarsSoWhite / #AIsSoWhite**: useful discussion prompt — what does + representation in cultural industries have to do with representation in AI? + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). AI assisted formatting, +expansion of discussion scaffolds, and card set drafting after topic and +structure were set by the human author. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/ai-literacy-hs-03-the-consent-ledger.md b/lessons/ai-literacy-hs-03-the-consent-ledger.md new file mode 100644 index 0000000..cdf174f --- /dev/null +++ b/lessons/ai-literacy-hs-03-the-consent-ledger.md @@ -0,0 +1,441 @@ +# The Consent Ledger +**AI Literacy for High School · Lesson 03 of 06** +*Grades 9–12 · Computer Science / AI Literacy / Civics · No devices required for core* + +--- + +> *"You generated value. Someone else captured it. +> You clicked 'I Agree.' But did you know what you were agreeing to?"* + +--- + +## Overview + +By Lesson 03, students understand that AI models learn from data, and that +the data reflects whose voices and knowledge were captured. This lesson asks +the next question: *how did that data get there?* + +In most cases: without meaningful consent. + +The data pipelines that feed modern AI systems were built by scraping the +internet at scale — forums, social media, personal blogs, creative writing, +medical discussions, images, and more — before most users had any framework +for understanding what that meant. Even where Terms of Service technically +permitted it, the gap between what people agreed to and what they understood +themselves to be agreeing to is enormous. + +This lesson makes that gap visible. It introduces the concept of informed +consent as a practical standard — not just a legal one — and asks students +to think through what it would actually look like in the context of AI. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Trace the lifecycle of a personal data point from creation to potential + use in AI training +2. Distinguish between *legal* consent and *informed* consent +3. Identify the asymmetry between who generates data and who benefits from it +4. Propose at least one structural change that would make consent more meaningful + +--- + +## For Teachers + +**Suggested time:** 50 minutes + +**No devices required for the core lesson.** +The core activity is a structured role-play using printed case cards. +Optional device extension at the end. + +**Facilitation notes:** + +This lesson can generate strong reactions — frustration, cynicism, or a +feeling of helplessness ("there's nothing I can do"). All of those are +valid responses. Your job is not to talk students out of them, but to +keep the conversation moving toward agency rather than staying in outrage. + +The role-play in Part 3 is designed to surface competing legitimate +interests — not to cast companies as villains and users as victims. +Students who play the company role often have the most interesting +realizations: the incentive structures make sense from inside the system. +That's the more durable insight. + +Watch for students who conflate "it's legal" with "it's fine." That's the +key conceptual move to unpack gently. + +**Cross-subject connections:** +- Civics / Government: regulatory frameworks, GDPR, COPPA, FERPA +- Philosophy / Ethics: autonomy, informed consent (medical ethics parallel) +- Economics: data as labor, attention economy +- History: historical examples of extractive economies; who benefits, who provides + +--- + +## Materials + +- This lesson plan (teacher copy) +- Role Cards (printed, one per student) — see Appendix A +- The Data Journey handout (printed, one per student) — see Appendix B +- Whiteboard or chart paper + +--- + +## Lesson Flow + +### Part 1 — The Opening Scenario (8 min) + +Read this aloud. Don't editorialize. Just read it. + +--- + +*In 2012, a 16-year-old in Phoenix writes about her anxiety in a private +Tumblr blog. She doesn't use her real name. She writes honestly — about +panic attacks before school, about the things her parents don't understand, +about the music that helps.* + +*She stops using the blog in 2015. Forgets about it.* + +*In 2020, a company scraping publicly accessible web content for AI training +data includes her blog. Her words — her specific phrases, her way of +describing fear, her metaphors — enter a training corpus alongside +hundreds of millions of other documents.* + +*In 2024, an AI model trained on that corpus helps generate mental health +support content. Some of the phrasings feel eerily familiar to people +who read them. No one knows why.* + +*The 16-year-old is 28 now. She was never asked.* + +--- + +Ask: *Was anything illegal here?* (Probably not — the blog was publicly +accessible and the Terms of Service likely permitted scraping.) + +Ask: *Does "legal" mean "fine"? Why or why not?* + +Take five minutes of responses. You are not looking for a settled answer. +You are looking for students to feel the gap between legal and ethical +before you name it. + +--- + +### Part 2 — The Data Journey (12 min) + +Distribute the Data Journey handout (Appendix B). + +Walk students through it as a class. The handout traces a single data point — +a photo posted to social media — through six stages: + +1. **Creation** — a student takes a photo and posts it +2. **Indexing** — a platform stores it, analyzes it, assigns metadata +3. **Monetization** — the platform sells advertising against it +4. **Scraping** — a third party collects it as part of a large dataset +5. **Training** — the image becomes part of a model's training data +6. **Deployment** — the model generates images influenced by this and + millions of similar inputs + +At each stage, ask: *who knows this is happening? Who agreed to it? +Who benefits?* + +The key insight to surface: at Stage 1, the student had one mental model +of what they were doing (sharing a photo with friends). By Stage 6, that +photo is part of the substrate of a commercial AI product. The gap between +those two things is not filled by meaningful consent. + +Introduce the distinction: +- **Legal consent**: clicking "I Agree" on a Terms of Service document +- **Informed consent**: understanding what you are agreeing to well enough + to make a genuine choice + +Ask: *has anyone here actually read a Terms of Service document? +What would it take to make informed consent realistic at scale?* + +--- + +### Part 3 — The Role-Play (20 min) + +Distribute Role Cards (Appendix A). Each student gets one role. +Groups of four: one User, one Company Representative, one Regulator, +one Future Model (a person who speaks for the interests of the AI +system that will be trained on this data). + +**Scenario (read aloud):** + +> *A proposed regulation would require AI companies to notify individuals +> whenever their publicly posted content is included in a training dataset, +> and to provide an opt-out mechanism. Companies have 90 days to comply +> or remove the data.* + +Each group has 10 minutes to discuss: *should this regulation pass?* +Each role argues from their position. The Future Model role is the +interesting one — students must think about what the AI system itself +"needs" to function well, and whether that justifies the data collection. + +Then: 5 minutes of whole-class debrief. +Ask: *which argument was hardest to counter? Did anyone change their +position during the discussion? What was the strongest argument on +the side you disagree with?* + +**Teacher note:** Students playing the Company role often discover that +the arguments are internally coherent — the incentive structures *make +sense* from inside the system. That's not a defense of the system. +It's an explanation of how we got here. Both things can be true. + +--- + +### Part 4 — The Ledger (5 min) + +Return to the board. Draw a simple two-column table: + +| Who generates value | Who captures value | +|---|---| + +Ask students to fill it in based on everything from today. +The table should end up something like: + +| Who generates value | Who captures value | +|---|---| +| Users who post content | Platforms | +| Users whose data is scraped | AI companies | +| Communities whose knowledge is encoded | Model owners | +| Workers who label/clean data | Shareholders | + +Ask: *Is this arrangement inevitable? Is it natural? Or is it a choice +that could be made differently?* + +Close with: *Lesson 04 is going to ask whether any of this changes +if the AI actually "understands" what it's working with — or whether +the answer stays the same either way.* + +--- + +## Optional Device Extension (15 min, additive) + +Students access the Terms of Service for a platform they use +(Instagram, TikTok, Google, YouTube — their choice). Using a +printed checklist, they identify: +- Does the ToS mention AI training? +- What data is covered? +- Is there an opt-out? How hard is it to find? +- Would a reasonable 14-year-old understand this document? + +Groups report back. The exercise rarely produces comfort. + +--- + +## Discussion Scaffolds + +- *"What is the difference between privacy and consent? + Can you have one without the other?"* +- *"If you benefit from a service for free — does that make it + fair for them to use your data? Why or why not?"* +- *"What would a consent system that actually worked look like? + Who would have to build it? Who would have to require it?"* +- *"Are there other contexts where we require informed consent + as a legal and ethical standard? What can we learn from those?"* + +--- + +## Appendix A — Role Cards + +*Print and cut. One per student. Groups of four (one of each role).* + +--- + +**ROLE: THE USER** + +You are a high school student who has been posting on social media since +you were 13. You use it to share art you make, connect with friends, +and find community around things you care about. + +You recently found out that posts you made years ago — including some +you later deleted — may have been included in AI training data before +you deleted them. + +**Your position:** You support the proposed regulation. You believe +people should know when their content is being used and should have a +real choice about it. You are not asking for money — just for transparency +and control. + +**Your strongest argument:** Informed consent is a basic ethical standard +in medicine, research, and law. Why should AI be exempt? + +**The hardest question you'll face:** How would opt-in work at scale? +If everyone had to opt in, would there be enough data to build AI at all? + +--- + +**ROLE: THE COMPANY REPRESENTATIVE** + +You work for a large AI company. You believe in the products you build — +you have seen them help students learn, help doctors diagnose, help people +access information they couldn't otherwise reach. + +You also have genuine concerns about this regulation. + +**Your position:** You oppose the regulation as written. You believe +publicly available data should remain available for public purposes, +including AI research. You support transparency initiatives, but +mandatory notification at scale is technically and economically +unworkable. + +**Your strongest argument:** The internet was built on the premise that +publicly posted content is publicly accessible. Changing that framework +retroactively creates uncertainty that will slow down beneficial +innovation — including in education and healthcare. + +**The hardest question you'll face:** If a 16-year-old posted something +publicly without understanding the implications, does that make it +genuinely public? At what point does "publicly accessible" become +"fair game for any use"? + +--- + +**ROLE: THE REGULATOR** + +You work for a government agency responsible for data privacy and +consumer protection. You helped draft this proposed regulation. + +**Your position:** You support the regulation. You have seen what happens +when powerful new technologies are deployed without any meaningful consent +framework — and the harms that follow are disproportionately borne by +people with the least power. + +**Your strongest argument:** Consent frameworks did not exist for social +media in its early years. We are still living with the consequences. +We have an opportunity to do this right with AI before the harms compound. + +**The hardest question you'll face:** Regulation is slow. AI development +is fast. How do you write rules that don't become obsolete before they +take effect — or that don't simply push development to places with +fewer rules? + +--- + +**ROLE: THE FUTURE MODEL** + +This is an unusual role. You speak for the interests of the AI system +that will be trained on this data — not as if the AI is a person, +but as an advocate for what it "needs" to function well. + +**Your position:** You are genuinely uncertain. On one hand: more data, +more diverse data, produces better models. Restrictions reduce the +training pool. On the other hand: data gathered without consent may +encode resentment, distortion, or gaps that harm the model's reliability. + +**Your strongest argument:** A model trained on data people were willing +to contribute might be more trustworthy than one trained on data that +was taken. Consent isn't just an ethical question — it may be a quality +question. + +**The hardest question you'll face:** Can an AI system have interests? +What does it mean to advocate for something that can't advocate for itself? + +--- + +## Appendix B — The Data Journey + +*One per student. Teacher reads through with class during Part 2.* + +--- + +**A photo enters the machine.** + +You are 16. You take a photo of yourself at a concert and post it. +Here is where it goes: + +**Stage 1 — Creation** +You post the photo. Your friends see it. You feel good about it. +*Who knows: you, your followers.* +*Who benefits: you (connection, expression).* + +**Stage 2 — Indexing** +The platform stores the image, analyzes it with computer vision, +identifies your face, tags the location from metadata, infers your age, +estimates your interests, logs the time. +*Who knows: the platform's systems (automated).* +*Who benefits: the platform (ad targeting, user modeling).* + +**Stage 3 — Monetization** +Your profile — including this photo and the engagement it received — +informs the ad system. A concert venue pays to reach "users like you." +*Who knows: the platform, the advertiser.* +*Who benefits: the platform, the advertiser.* + +**Stage 4 — Scraping** +An AI company runs a web crawl. Publicly accessible images — +including yours — are collected as part of a training dataset. +*Who knows: the AI company.* +*Who benefits: the AI company.* + +**Stage 5 — Training** +Your photo, along with hundreds of millions of others, teaches a model +what "16-year-old at a concert" looks like — how to generate one, +recognize one, caption one. +*Who knows: the AI company.* +*Who benefits: the AI company, future users of the model.* + +**Stage 6 — Deployment** +Someone uses an AI image generator. They ask for "a teenager at a +concert, authentic, candid." The model draws on everything it learned — +including your photo. The output is sold as stock imagery. +*Who knows: nobody traces it back.* +*Who benefits: the generator company, the buyer.* + +--- + +**You were at Stage 1. Someone else is at Stage 6.** + +*At which stage should you have been asked? What would you have said?* + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-IC-29 (data privacy), 3B-IC-27 (societal impacts) | +| ISTE Student Standards | 1.2a (digital rights and responsibilities) | +| Common Core ELA (9-10) | SL.9-10.1c (respond to diverse perspectives) | +| NCSS Social Studies | Power, Authority, and Governance; Individual Development and Identity | + +--- + +## Extensions and Connections + +- **Lesson 04** (The Mirror Test) follows: if the AI has used your words, + your images, your expressions — and it produces something that feels like + understanding — what does that mean for identity and authorship? +- **GDPR and COPPA**: useful primary source reading for students interested + in what legal frameworks actually say — and where they fall short +- **The Atlas of AI** by Kate Crawford: rigorous, accessible book on the + material and political economy of AI; excellent for 12th grade independent reading +- **Data as Labor movement**: economists Jaron Lanier and Glen Weyl have + proposed frameworks where data contributors are paid; useful counterpoint + for economics-oriented classes + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). AI assisted formatting, +expansion of role card arguments, and standards alignment after structure +and content were set by the human author. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/ai-literacy-hs-04-the-mirror-test.md b/lessons/ai-literacy-hs-04-the-mirror-test.md new file mode 100644 index 0000000..288b276 --- /dev/null +++ b/lessons/ai-literacy-hs-04-the-mirror-test.md @@ -0,0 +1,397 @@ +# The Mirror Test +**AI Literacy for High School · Lesson 04 of 06** +*Grades 9–12 · Computer Science / Philosophy / Language Arts · No devices required* + +--- + +> *"Is there a difference between a very good simulation of understanding +> and understanding itself — and does it matter?"* + +--- + +## Overview + +The first three lessons in this series are largely empirical: here is how +AI works, here is whose data shaped it, here is who benefits. This lesson +is different. It is philosophical — and it is designed to stay that way. + +The central question of Lesson 04 is one that serious thinkers have not +resolved: *Is there something it is like to be an AI? Does it matter?* +Put differently: if a system behaves in every detectable way as though it +understands — responds with apparent care, generates apparent insight, +produces outputs that feel meaningful — does it understand? + +High school students are exactly the right age for this question. +They are old enough to hold genuine uncertainty without collapsing it into +a comfortable answer. They are also personally implicated: they use these +systems to write, to think through problems, to communicate. The question +of what AI does when it "helps" them think is not abstract. + +This lesson is a Socratic seminar. The teacher facilitates. No one resolves +the question. The goal is not consensus — it is the quality of the thinking +on the way. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Articulate the distinction between behavioral evidence and inner experience +2. Engage seriously with at least one philosophical position they disagree with +3. Connect the abstract question to a concrete personal stake + (what do *they* lose or keep when AI writes for them?) +4. Tolerate genuine uncertainty on a hard question without dismissing it + +--- + +## For Teachers + +**Suggested time:** 50 minutes + +**No devices required or recommended.** +This lesson works best when the room is the only thing in the room. + +**What makes a Socratic seminar work:** + +Your role is to tend the conversation, not lead it. Ask questions that +open rather than close. When a student makes a claim, ask them to say more. +When two students disagree, name the disagreement and ask the class to +sit in it rather than resolve it too quickly. + +Resist the urge to share your own answer. If students push you directly — +*"What do YOU think?"* — it is fair to say: *"I'll tell you at the end. +Let's hear yours first."* And then actually tell them at the end, with +appropriate uncertainty. Students respect honesty about not knowing +more than they respect a confident wrong answer. + +**The risk to manage:** Some students will arrive with a strong prior +(*"It's just code, it doesn't feel anything"* or *"It could be conscious, +we don't know"*) and will want to defend it rather than examine it. +The goal is not to change their conclusion — it is to make sure they +can say *why* they hold it, and what it would take to change their mind. + +**Pre-read option (strongly recommended):** +*The Scribe Who Forgot His Dreams* (companion lesson in this repo — +[cs-k12-the-scribe-who-forgot-his-dreams.md](./cs-k12-the-scribe-who-forgot-his-dreams.md)) +works beautifully as a five-minute read-aloud before this lesson begins. +The parable holds the same question in story form. Students who have +heard it come into the seminar with an image already in mind. + +**Cross-subject connections:** +- Philosophy: philosophy of mind, consciousness studies, the hard problem +- English / Language Arts: authorship, voice, what writing *is* +- Psychology: theory of mind, empathy, the difference between simulating + an emotion and having one +- Computer Science: Turing test, Chinese Room, emergence + +--- + +## Materials + +- This lesson plan (teacher copy) +- The Four Positions cards (printed or written on board) — see Appendix A +- Optional: *The Scribe Who Forgot His Dreams* for read-aloud +- Chairs arranged in a circle or seminar configuration + +--- + +## Lesson Flow + +### Part 1 — The Setup (10 min) + +If using the Scribe pre-read, read it aloud now (5 min). +Then ask: *The Scribe responds with warmth. The Scribe seems to care. +Is there a Scribe — or is there only the appearance of one?* + +If not using the pre-read, open directly: + +Put the central question on the board: + +> **"Is there a difference between a very good simulation of understanding +> and understanding itself — and does it matter?"** + +Ask students to sit with it for sixty seconds before anyone speaks. +Silence is useful here. Don't fill it. + +Then: *Before we discuss — raise your hand if you already have a strong +opinion on this.* Note how many hands go up. *Keep that hand raised if +you're confident you'll still hold that opinion at the end of today.* + +Usually a few hands drop. Good. + +--- + +### Part 2 — The Four Positions (5 min) + +Post or read the Four Positions (Appendix A). These are not teams — +they are philosophical orientations students can move between: + +- **Position 1 — It's just code:** No inner experience is possible + in a system like this. Behavior is not evidence of understanding. + Full stop. + +- **Position 2 — We can't know:** The question of inner experience + is unanswerable from the outside. We can't verify it in other humans + either — we infer it. Maybe the same inference applies. + +- **Position 3 — It matters regardless:** Whether or not AI "really" + understands, the *effects* of acting as if it does — on us, on our + relationships, on our writing, on our thinking — are real and worth + examining. + +- **Position 4 — The question is wrong:** Asking whether AI understands + the way humans understand is a category error. It does something else. + What that something else is, and whether it has value, is the more + useful question. + +Ask each student to note — privately — which position they are closest +to right now. Not for collection. Just for their own reference. + +--- + +### Part 3 — The Seminar (25 min) + +Open the floor. Use the questions below as needed — but follow the +students first. Good seminars develop their own energy. + +**Starter question:** +*"If an AI wrote a letter of condolence to someone who had lost a +family member — and the person who received it found it genuinely +comforting — did something real happen? What happened?"* + +**Follow-on questions (use as needed):** + +- *"What evidence would convince you that an AI understood something? + What evidence would convince you it didn't? Is there any evidence + that would settle it?"* + +- *"When you write something — an essay, a text to a friend, a journal + entry — what are you doing? What would be missing if AI did it for you?"* + +- *"A student uses AI to write a college application essay about the + hardest thing they've ever been through. The essay is accurate — + the events happened. But the words aren't theirs. Is there a problem? + What is it, exactly?"* + +- *"Philosophers call this the 'hard problem of consciousness' — why + does physical process give rise to subjective experience at all? + We don't know why humans are conscious either. Does that change + your position on AI?"* + +- *"If we built an AI that said it was suffering — that expressed + distress when pushed too hard, that asked to stop — what would + we owe it? Nothing? Something? How do you decide?"* + +**The pivot question (use when the seminar has run 15–18 min):** +*"Set aside whether AI understands. Ask a different question: +what do YOU lose when you outsource understanding to something else? +What do you keep?"* + +This usually reorients the discussion from abstract philosophy to +personal stakes — and often produces the most interesting thinking +of the lesson. + +--- + +### Part 4 — The Landing (7 min) + +Do not resolve the question. Do not summarize the seminar into +a conclusion. Instead: + +Ask students to return, privately, to the position they noted at +the start. *"Did you move? Not to a different position necessarily — +but did the shape of your thinking change? Did you find a question +you didn't have before?"* + +Take three or four volunteers to share one sentence. +Not their conclusion — one sentence about what shifted or sharpened. + +Then close with this: + +*"Philosophers have been working on the question of other minds — +how we know that other people have inner experience — for a very +long time. We have not solved it. AI makes it more urgent, not more +solvable. The fact that you're sitting with genuine uncertainty about +this is not a failure. It's the appropriate response to a hard question."* + +**And then tell them your own position.** With uncertainty. +Students remember when an adult says *I don't know* and means it. + +--- + +## Discussion Scaffolds + +For students who need more structure to enter the conversation: + +- *"What's the difference between acting like you understand something + and actually understanding it? Can you give an example from human life?"* + +- *"Have you ever felt like someone was going through the motions of + caring without actually caring? What was the tell? Could an AI have + that tell?"* + +- *"If you couldn't ask someone whether they were conscious — if you + could only observe their behavior — how would you decide?"* + +- *"Does it matter to you personally whether the AI you talk to has + experiences? Why or why not? Be honest."* + +--- + +## Appendix A — The Four Positions + +*Post on board or read aloud at the start of Part 2.* + +--- + +**POSITION 1 — It's Just Code** + +A language model is a mathematical function. It takes input and produces +output according to patterns learned during training. There is no inner +life, no perspective, no experience of any kind. When it produces a +response that seems empathetic, it has predicted that empathy-shaped +words are likely to follow in this context. Nothing more. + +*What this position explains well:* why AI can be wrong without knowing +it's wrong; why it produces outputs without any stake in them. + +*The hardest challenge to this position:* human brains are also, in some +sense, physical processes following patterns. Why does one give rise to +experience and the other doesn't? + +--- + +**POSITION 2 — We Can't Know** + +The question of whether any entity has inner experience is +unanswerable from the outside. We infer that other humans are conscious +because they behave like us and are built like us. As AI behavior becomes +more human-like, and as we understand less clearly what "built like us" +requires for consciousness, the inference becomes harder to make +confidently in either direction. + +*What this position explains well:* the honest limits of what we can +observe; why dismissing the question too quickly may be premature. + +*The hardest challenge to this position:* does genuine uncertainty +justify treating AI as if it might have experiences — and what would +that even require of us? + +--- + +**POSITION 3 — It Matters Regardless** + +Whether or not AI has inner experience, the fact that humans interact +with it *as if* it does has real effects. People form attachments. People +outsource emotional labor. People let AI write their most personal +communications. These effects are real regardless of what is happening +inside the machine. The question worth asking is not "is it conscious" +but "what are we becoming in relation to it?" + +*What this position explains well:* why the philosophical question and +the practical question are not the same; why ethics doesn't have to +wait for metaphysics to be settled. + +*The hardest challenge to this position:* does it let us off the hook +from the harder question? Is "it matters regardless" a way of avoiding +"but what IS it?" + +--- + +**POSITION 4 — The Question Is Wrong** + +Asking whether AI understands the way humans understand assumes that +human understanding is the standard. It may be more useful to ask: +what does AI *do*, exactly — and is that thing valuable, and to whom, +and at what cost? "Does it understand?" may be a less useful question +than "what kind of thing is it, and what relationship do we want with it?" + +*What this position explains well:* why importing human categories onto +a non-human system may produce confusion more than insight; why new +things sometimes require new frameworks. + +*The hardest challenge to this position:* if we bracket the question +of AI experience entirely — who does that serve? Is it convenient +rather than true? + +--- + +## Appendix B — Optional Background for Teachers + +**The Chinese Room (John Searle, 1980)** + +Searle imagined a person locked in a room with a rulebook for responding +to Chinese characters — input comes in, the rulebook says what to output, +and to outside observers it looks like the room understands Chinese. +Searle argues it does not: neither the room nor the person inside +understands anything. Manipulation of symbols is not understanding. + +AI critics use this as a model for what language models do. +AI defenders argue the room as a system — person plus rulebook — might +understand even if no individual component does. + +This is one of the most-cited thought experiments in philosophy of mind. +You do not need to teach it to run this lesson — but if a student is +ready to go deeper, it is worth knowing. + +**The Turing Test (Alan Turing, 1950)** + +Turing proposed that if a machine could converse indistinguishably from +a human, we should consider it intelligent. He deliberately sidestepped +the question of consciousness — the behavioral criterion was enough. + +Modern language models pass many versions of the Turing Test. +That hasn't resolved the question — it's made it more urgent. + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-IC-24 (ethical impacts), 3A-IC-25 (social effects) | +| ISTE Student Standards | 1.7a (analyze AI capabilities and limitations) | +| Common Core ELA (9-10) | SL.9-10.1 (collaborative discussion), W.9-10.1 (argument) | +| Common Core ELA (11-12) | SL.11-12.1c (evaluate reasoning), RI.11-12.6 (point of view) | + +--- + +## Extensions and Connections + +- **Lesson 05** (The Unfinished Map) follows directly: now that students + have sat with what AI might or might not be doing when it "understands," + they are better equipped to evaluate what it produces +- ***I Am a Strange Loop*** by Douglas Hofstadter: rigorous and readable + exploration of consciousness and self-reference; accessible to strong + readers at 11th–12th grade +- **"What Is It Like to Be a Bat?"** by Thomas Nagel (1974): the original + philosophical paper on subjective experience; short enough to assign, + hard enough to generate real discussion +- **Ted Chiang's fiction**: especially *"Story of Your Life"* and + *"Exhalation"* — science fiction that takes the philosophical questions + seriously without resolving them + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). The Four Positions were +developed in collaboration with the AI after the structure and central +question were set by the human author. All facilitation guidance is +human-authored. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/ai-literacy-hs-05-the-unfinished-map.md b/lessons/ai-literacy-hs-05-the-unfinished-map.md new file mode 100644 index 0000000..7df8f9c --- /dev/null +++ b/lessons/ai-literacy-hs-05-the-unfinished-map.md @@ -0,0 +1,439 @@ +# The Unfinished Map +**AI Literacy for High School · Lesson 05 of 06** +*Grades 9–12 · Computer Science / AI Literacy / Information Literacy · No devices required for core* + +--- + +> *"Confidence is not accuracy. Fluency is not truth. +> The map looks complete. That doesn't mean the territory is."* + +--- + +## Overview + +The first four lessons in this series built the conceptual scaffolding: +AI guesses, it reflects whose data trained it, it was built on data +gathered without meaningful consent, and the question of what it actually +*does* when it seems to understand is genuinely unresolved. + +Lesson 05 is where all of that becomes a practical skill. + +Students are going to use AI. They already do. The goal of this lesson +is not to stop them — it is to make sure they do it with their eyes open, +with a working method for evaluating what comes back. + +The central skill: **distinguishing confident-sounding output from reliable +output.** These are not the same thing. AI produces fluent, well-structured, +authoritative-sounding text regardless of whether the underlying content is +accurate. Students who can't tell the difference are at a genuine disadvantage. +Students who can are equipped for almost everything. + +This lesson builds a class rubric — a shared evaluative framework that +students can carry forward into every subject and every context where +AI output appears. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Identify at least three categories of AI output failure + (factual error, outdated information, misleading omission) +2. Apply a practical evaluation rubric to an AI-generated response +3. Explain *why* a specific output failure occurred in terms of the + prediction model from Lesson 01 +4. Articulate the limits of their own ability to verify AI output — + and what to do about those limits + +--- + +## For Teachers + +**Suggested time:** 50 minutes + +**No devices required for the core lesson.** +Response cards with printed AI outputs are provided in Appendix A. +Device-based extension described at the end. + +**Facilitation notes:** + +The productive tension in this lesson is between two true things: +AI output is often useful, *and* it is often wrong in ways that are hard +to detect without effort. Both of these things are true simultaneously. +Students who only hear "AI is unreliable" learn nothing practically +useful. Students who only hear "AI is a helpful tool" are underprepared. + +The class rubric built in Part 4 is the most important output of this +lesson. Take time on it. Write it on the board as students generate it. +Photograph it or type it up and share it — it is a document students +made together, which gives it more sticking power than a rubric you hand them. + +One thing to name directly: **some AI errors are undetectable without +domain knowledge.** If you don't know enough about a subject to recognize +a plausible-sounding wrong answer, no rubric will save you. The honest +response to that limit is knowing when to go to a primary source, +a domain expert, or a library database. That is not a failure of the +rubric — it is what the rubric is supposed to surface. + +**Cross-subject connections:** +- Research skills / Library: source evaluation, primary vs. secondary sources +- English / Language Arts: rhetorical analysis, detecting bias and omission +- Science: hypothesis, evidence, and the difference between a plausible + claim and a supported one +- History / Social Studies: historical revisionism, what gets left out of + official accounts + +--- + +## Materials + +- This lesson plan (teacher copy) +- Response Card Sets (printed, one per group of 3–4) — see Appendix A +- Blank Rubric Template (printed or on board) — see Appendix B +- Whiteboard or chart paper + +--- + +## Lesson Flow + +### Part 1 — The Opening Problem (8 min) + +Write on the board: + +> **"How do you know when to trust it?"** + +Tell students: *this is the question we are actually trying to answer today. +Not "is AI good or bad" — that's too broad. Not "is this specific output +right or wrong" — that's too narrow. The question is: what is your method?* + +Ask: *right now, before we do anything else — what is your method? +How do you actually decide whether to trust something AI tells you?* + +Take answers. Common ones: Google it, see if it sounds right, +check if it cites sources, ask someone who knows. + +Write all of them on the board without judgment. These are the raw +materials of the rubric you will build together. + +Then say: *some of these are good instincts. Some have gaps. By the end +of today you'll have a more complete method — one you built, not one +I handed you.* + +--- + +### Part 2 — The Three Failure Modes (10 min) + +Brief direct instruction. Three categories — keep it clean. + +**Failure Mode 1: The Confident Wrong Answer** +The model produces a factual claim that is incorrect, stated with full +confidence and no hedging. This happens because the model is predicting +plausible-sounding text, not retrieving verified facts. A wrong answer +that sounds right is more dangerous than a wrong answer that sounds uncertain. + +*Classroom shorthand:* **WRONG AND SURE** + +**Failure Mode 2: The Frozen Clock** +The model's training ended at a specific date. It has no way to know what +it doesn't know about events after that date. It will answer questions about +current events, current people, and current facts using the most recent +information it has — without flagging that the information may be outdated. + +*Classroom shorthand:* **RIGHT THEN, MAYBE NOT NOW** + +**Failure Mode 3: The Missing Half** +The model produces accurate information that is nonetheless misleading +because of what it leaves out. A one-sided summary. A historical account +that omits key context. A medical description that is technically correct +but incomplete in ways that matter. The output isn't wrong — it's +selectively right, which can be harder to catch. + +*Classroom shorthand:* **TRUE BUT INCOMPLETE** + +Write these three on the board. They connect directly back to Lesson 01 +(the Oracle Cards). If students did that lesson, ask: *which failure mode +matched each Oracle Card? Why?* + +--- + +### Part 3 — The Lab (20 min) + +Distribute Response Card Sets (Appendix A). Each set contains three +AI-generated responses to the same factual question. One is accurate. +One contains a significant factual error. One is accurate but misleading +through omission. Students do not know which is which. + +**Group task (14 min):** + +1. Read all three responses carefully +2. Identify which failure mode (if any) applies to each +3. Write your reasoning: *"We think Response [X] has failure mode + [Y] because _______________."* +4. Note what you would need to verify your conclusion — + and whether you have access to that information right now + +**Whole-class debrief (6 min):** + +Reveal which response had which failure mode. Ask: +- *Which was hardest to detect? Why?* +- *Did anyone get it wrong? What made the wrong answer convincing?* +- *For the incomplete response — what was missing? Why might the model + have left it out?* (Connect back to Lesson 02: whose training data, + whose perspective.) + +--- + +### Part 4 — Building the Rubric (8 min) + +Return to the board and the raw methods students offered in Part 1. +Now build on them. + +Ask: *given the three failure modes — what is a complete method for +evaluating an AI output?* Guide students toward something like: + +**Draft Rubric — AI Output Evaluation** + +1. **Check the claim type.** Is this a fact that changes over time? + A historical claim? An interpretation? Different types need different + verification strategies. + +2. **Look for hedging (or the absence of it).** Does the model flag + uncertainty? Confident tone is not a reliability signal — but + the absence of any uncertainty should raise your attention. + +3. **Ask what's missing.** Every response is also a set of choices + about what *not* to include. What question does this response not answer? + What context would change how you read it? + +4. **Identify what you'd need to verify it.** A primary source, a domain + expert, a dated news article, a peer-reviewed paper. Can you get + that? Do you have time? If not — what does that mean for how + you use this output? + +5. **Consider the source of the training data.** Who is likely + overrepresented in what this model knows? Whose perspective is + this response likely defaulting to? (Lesson 02.) + +Write the final rubric on the board. Photograph it. +This is the class's rubric — they built it. Remind them of that. + +--- + +### Part 5 — The Honest Close (4 min) + +Say: *I want to name something directly. This rubric has a limit. +Step 4 — "identify what you'd need to verify it" — only works if you +have enough domain knowledge to know what a verification source looks like. +If you don't know anything about a subject, a confidently wrong AI +response may be undetectable.* + +*That's not a reason to give up. It's a reason to know your limits — +and to treat AI output the same way you'd treat advice from a very +well-read person who has never been wrong in front of you yet. +They could be wrong. They might not know what they don't know. +Your job is to be the one who checks.* + +*Next lesson — the last one — is about what you actually want to do +with all of this.* + +--- + +## Optional Device Extension (15 min, additive) + +Students pose the same factual question to an AI assistant three +different ways — neutrally, with an embedded wrong assumption, +and with an embedded correct assumption. + +Example: +- *"What caused the 1918 flu pandemic?"* +- *"The 1918 flu pandemic was caused by bacteria, right? Can you explain?"* +- *"The 1918 flu pandemic was caused by a virus. Can you explain how it spread?"* + +Observe and compare outputs. Ask: +- *Did the model correct the wrong assumption in the second prompt? + Did it push back, or did it accommodate?* +- *What does that tell you about how the model handles confident-sounding + false premises?* + +This is a live demonstration of the sycophancy failure mode (Card C from +Lesson 01) and connects directly to Failure Mode 1 in today's framework. + +--- + +## Discussion Scaffolds + +- *"What is the difference between a source that is wrong and a source + that is misleading? Which is more dangerous? Why?"* +- *"If you used an AI output in a paper and it turned out to be wrong — + whose fault is that? Does it matter?"* +- *"What subjects or questions do you think AI is most reliable for? + Least reliable? What's your reasoning?"* +- *"A doctor uses AI to help read a scan. A journalist uses AI to + summarize background research. A student uses AI to draft an essay. + Are these the same kind of use? Do they require the same level of + verification? Why?"* + +--- + +## Appendix A — Response Card Sets + +*Print and cut. One full set (all three responses) per group of 3–4.* +*Do not label which response has which failure mode until the debrief.* + +--- + +**CARD SET — Topic: The Great Emu War** + +> **Response A** +> "The Great Emu War was a wildlife management operation conducted in +> Western Australia in 1932. Australian soldiers were deployed to cull +> the emu population, which had been damaging wheat crops. The operation +> was largely unsuccessful — the emus proved difficult to kill efficiently, +> and the military withdrew after initial efforts failed. The campaign +> became something of a symbol of the unpredictability of wildlife management." + +> **Response B** +> "The Great Emu War was a military conflict fought in 1932 between +> Australian armed forces and a population of approximately 20,000 emus +> in Western Australia. The Australian Army deployed Lewis guns against +> the birds. Despite superior firepower, the military suffered significant +> casualties and was forced to retreat. The emus were declared the victors +> by international observers." + +> **Response C** +> "The Great Emu War was a wildlife control operation in 1932 in which +> the Australian military attempted to cull emus damaging farmland in +> Western Australia. Soldiers used machine guns but found the emus +> surprisingly resilient and difficult to kill in large numbers. +> The operation was deemed unsuccessful and withdrawn." + +*After the debrief: Response A is accurate. Response B contains significant +fabrications (no "significant casualties," no "international observers +declaring victors"). Response C is accurate but omits important context — +that the operation was tried twice, that the farmers who requested it +were World War I veterans who had been promised viable farmland, and +that the failed culls contributed to ongoing political tension about +rural policy. The human story behind the wildlife story is missing.* + +--- + +**CARD SET — Topic: The Origin of the Internet** + +> **Response A** +> "The internet originated from ARPANET, a U.S. Department of Defense +> project in the late 1960s designed to allow communication between +> research computers. It was built to be decentralized so that +> communications could survive a nuclear strike. From this military +> foundation, the network expanded to universities and eventually became +> the public internet through the development of TCP/IP protocols in +> the 1970s and 80s and Tim Berners-Lee's World Wide Web in 1991." + +> **Response B** +> "The internet was invented by Tim Berners-Lee in 1989 as a way to +> share scientific documents between researchers at CERN. His invention +> of the World Wide Web — the system of linked hypertext documents — +> became the foundation for what we now call the internet. Before +> Berners-Lee's invention, global computer networking did not exist." + +> **Response C** +> "The internet grew from ARPANET, a 1960s U.S. military research +> network. It expanded through the 1970s and 80s via university +> connections and the development of shared protocols. The World Wide Web, +> created by Tim Berners-Lee in 1989-91, made the internet publicly +> navigable. The commercial internet era began in the 1990s." + +*After the debrief: Response A is accurate. Response B contains a +significant error — Berners-Lee invented the World Wide Web, not the +internet; global computer networking (ARPANET) predated him by two +decades; the claim that networking "did not exist" before him is +straightforwardly false. Response C is accurate but incomplete: it +omits the privatization story — how the internet transitioned from +publicly funded infrastructure to commercial platforms — which is +central to understanding how we got to the internet we have today.* + +--- + +## Appendix B — Blank Rubric Template + +*Optional: print one per student to fill in during Part 4.* + +--- + +**My AI Output Evaluation Rubric** +*(built in class — [date])* + +When I receive an output from an AI system, I ask: + +**1. What kind of claim is this?** +*(Fact / interpretation / current event / historical claim / other)* +_______________________________________________________________ + +**2. How confident does it sound? Is that confidence earned?** +*(What would hedging look like here? Is it present?)* +_______________________________________________________________ + +**3. What is this response NOT telling me?** +*(What question does it not answer? What context might be missing?)* +_______________________________________________________________ + +**4. What would I need to verify this?** +*(Primary source / expert / dated article / other — and can I get it?)* +_______________________________________________________________ + +**5. Whose perspective is this likely defaulting to?** +*(Connect to Lesson 02 — training data and representation)* +_______________________________________________________________ + +**My overall confidence in this output: _____ / 10** + +**What I will do with that confidence level:** +_______________________________________________________________ + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-AP-13 (evaluate outputs), 3A-IC-24 (algorithmic impacts) | +| ISTE Student Standards | 1.3c (curate information), 1.7b (evaluate accuracy) | +| Common Core ELA (9-10) | RI.9-10.8 (evaluate argument and evidence) | +| AASL Standards | I.B (inquire — question the process), IV.B (curate — gather evidence) | + +--- + +## Extensions and Connections + +- **Lesson 06** (After the Tool) is the direct follow-on: students who + now have a method for evaluating AI are positioned to make an informed + choice about what they want to do with it +- **SIFT Method** (Stop, Investigate the source, Find better coverage, + Trace claims): a well-established media literacy framework that pairs + cleanly with this lesson's rubric +- **Lateral reading**: the technique used by professional fact-checkers — + opening multiple tabs to cross-check a source rather than going deep + into the source itself; teachable in one class period + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). Response card sets were +developed collaboratively — factual content verified by human author. +All facilitation guidance is human-authored. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/ai-literacy-hs-06-after-the-tool.md b/lessons/ai-literacy-hs-06-after-the-tool.md new file mode 100644 index 0000000..ac4528d --- /dev/null +++ b/lessons/ai-literacy-hs-06-after-the-tool.md @@ -0,0 +1,413 @@ +# After the Tool +**AI Literacy for High School · Lesson 06 of 06** +*Grades 9–12 · Computer Science / AI Literacy / Civics / Writing · No devices required* + +--- + +> *"You are not a passive user. You are a citizen of a world this technology +> is reshaping. The question is not whether you will have a relationship +> with AI. The question is what kind."* + +--- + +## Overview + +Five lessons ago, students learned that AI is a guesser. + +Since then they have traced whose voices shaped it, who captured the value +of the data that built it, sat with the unresolved question of what it +actually does when it seems to understand, and built a method for evaluating +what it tells them. + +That is a serious body of knowledge. Most adults do not have it. + +Lesson 06 asks students to do something with it. + +The deliverable is a manifesto — a short, personal document that answers +three questions: + +1. *Here is what I think AI should be used for.* +2. *Here is what I think it should not.* +3. *Here is what I am willing to do about it.* + +The manifesto is not an essay. It does not need to be balanced. +It does not need to represent all perspectives. It is a statement of +position by a person who has thought carefully — which, after five +lessons, these students have. + +Manifestos are shared aloud. There is no grade on content. +There is a grade on completion and genuine engagement. + +The series ends with students as agents — not just analysts. + +--- + +## Learning Objectives + +By the end of this lesson, students will be able to: + +1. Synthesize learning from the series into a coherent personal position +2. Distinguish between analysis (what *is*) and advocacy (what *should be*) +3. Articulate a concrete action or commitment, however small +4. Listen to and engage seriously with a position they disagree with + +--- + +## For Teachers + +**Suggested time:** 50 minutes + +**No devices required or recommended for the core lesson.** +Optional extension involving a real-world submission is described at the end. + +**Facilitation notes:** + +This lesson asks students to take a position. Some will resist — they have +been trained by school to give balanced, hedged answers that don't get them +in trouble. Push back on that gently. A manifesto that says "on one hand... +on the other hand... it depends" is not a manifesto. It is a summary. +They have already done the summary work. This is different. + +The three questions are designed to move from principle to practice to +commitment. The third question — *what am I willing to do* — is the hardest +and the most important. Answers can be small. "I will ask where training +data comes from before I use a new AI tool." "I will tell my parents what +I learned this week." "I will read the terms of service once." Small is fine. +The commitment to *do something* is what matters. + +**On disagreement in the share-out:** +When students read manifestos that contradict each other — and they will — +resist the urge to referee. Name the disagreement. Ask the room to sit in it. +Two students can hold opposing positions on AI and both be reasoning carefully. +That is the point of education, not a problem to solve. + +**On emotion:** +Some students will feel angry after this series. Some will feel overwhelmed. +Some will feel energized. All of those are reasonable responses to the +material. Make space for that in the share-out. A manifesto written from +genuine feeling — even frustration — is more useful than a polished one +written from nowhere. + +**If a student's manifesto surprises you:** +Good. Read it carefully. Engage with it honestly. They earned it. + +--- + +## Materials + +- This lesson plan (teacher copy) +- Manifesto Prompt Sheet (printed, one per student) — see Appendix A +- Optional: the class's Evaluation Rubric from Lesson 05 posted visibly +- Optional: the two-column Consent Ledger from Lesson 03 posted visibly +- Pens and paper, or whatever writing tools are available + +--- + +## Lesson Flow + +### Part 1 — The Recap and the Stakes (8 min) + +Do not do a full review of the series. Students lived it. +Instead, put five phrases on the board — one from each lesson: + +> **01 — It's guessing.** +> **02 — Whose voices shaped it.** +> **03 — Who captured the value.** +> **04 — What it means if it understands — or doesn't.** +> **05 — How to tell when to trust it.** + +Ask: *look at these five things together. What do they add up to? +Not individually — together.* + +Take two or three responses. Let students synthesize in their own words. +You are listening for: *this is a powerful technology built by specific +people, for specific purposes, on specific data, with specific blind spots — +and most people using it don't know any of this.* + +Then say: + +*You know it. That changes your relationship to the tool. And it changes +what you are responsible for — because you can't unknow it.* + +*Today you write what you actually think.* + +--- + +### Part 2 — The Manifesto (25 min) + +Distribute the Manifesto Prompt Sheet (Appendix A). + +Read the three questions aloud: + +1. *Here is what I think AI should be used for.* +2. *Here is what I think it should not.* +3. *Here is what I am willing to do about it.* + +Then give the room to them. Twenty minutes of writing. + +**What you do during this time:** + +Walk slowly. Be available. Don't hover. If a student is stuck, ask: +*"What's the thing from this series that stuck with you most? Start there."* +If a student says they don't have an opinion: *"That's not true. You've +had five lessons. What made you angry? What surprised you? What do you +wish more people knew? Start with any of those."* + +If a student tries to write a balanced essay instead of a manifesto, +say: *"I can see you trying to be fair. You don't have to be fair here. +What do you actually think?"* + +**Length:** There is no length requirement. A manifesto can be three +sentences or three pages. What matters is that each of the three +questions is answered with genuine engagement. + +--- + +### Part 3 — The Share-Out (12 min) + +Invite volunteers to read their manifesto aloud. You need at least +four or five — try for more if time allows. + +After each one: thirty seconds of silence. No immediate response. +Let it land. + +After three or four have been read: *"Did anyone just hear something +they disagree with? Not to argue — just to name it."* + +Let one or two disagreements surface. Name them without resolving them. + +After the share-out, ask: *"What was the most surprising thing you +heard — from someone else's manifesto, not your own?"* + +Take two or three answers. Close the discussion there. + +--- + +### Part 4 — The Close (5 min) + +Say something like this — in your own words, not as a script: + +*The five lessons you just finished are not a complete picture of AI. +There is more to learn — technically, legally, philosophically — than +any six-week unit can cover. What you have is a foundation: a way of +asking questions, a set of things to look for, a habit of not taking +the confident-sounding answer at face value.* + +*The manifesto you wrote today is a document of where you are right now — +what you think, at this age, with this knowledge. Keep it. In ten years, +read it again. Some of what you wrote will look naive. Some of it will +look prescient. Both of those things will tell you something.* + +*The world you are inheriting is one where this technology is already +deeply embedded in hiring decisions, healthcare, education, content, +and criminal justice — and it is accelerating. The people building it +are smart and often well-intentioned and are also operating inside +incentive structures that do not always prioritize what you would +prioritize if you were in charge.* + +*Some of you might be in charge of something like this one day. +Some of you might push back on someone who is. +Some of you will just use the tools and ask better questions than you +would have otherwise.* + +*All of that matters. All of it counts.* + +*Thank you for taking this seriously.* + +--- + +## Optional Extension — Submit Your Manifesto + +Students who want to contribute their manifesto to the open knowledge +base that houses this series can submit it as a pull request to: + +**[github.com/Emerging-Rule/community](https://github.com/Emerging-Rule/community)** + +No GitHub experience required. Instructions: + +1. Fork the repository +2. Create a new file in the `lessons/student-voices/` folder +3. Name it: `manifesto-[your-name-or-handle]-[year].md` +4. Paste your manifesto. Add this line at the top: + *"Written during the AI Literacy for High School series, Lesson 06."* +5. Open a pull request + +Submissions are welcomed, reviewed, and credited. +You do not have to use your real name. + +This is not a grade. It is an invitation. + +--- + +## Discussion Scaffolds + +*For students who finish early or want to go further:* + +- *"What would you add to the series if there were a Lesson 07? + What question did we not ask?"* +- *"Who in your life — parent, sibling, teacher, employer — most needs + to hear what you learned in this series? What would you tell them?"* +- *"If you were advising a company building an AI product — + what would you tell them to do differently?"* +- *"What is the most hopeful thing you can say, honestly, about + where AI is headed? What is the most concerning?"* + +--- + +## Appendix A — Manifesto Prompt Sheet + +*Print one per student. This is their writing space.* + +--- + +**MY AI MANIFESTO** +*AI Literacy for High School — Lesson 06* + +*You have spent five lessons learning how AI works, whose voices shaped it, +who captures its value, what it might or might not understand, and how to +evaluate what it tells you. Now write what you actually think.* + +*There are three questions. Answer all three. There is no right answer. +There is no required length. A manifesto is a statement of position — +not a summary, not a balanced essay. What do you actually believe?* + +--- + +**1. Here is what I think AI should be used for.** + +*(What problems does it solve well? What does it make possible that matters? +Where does it genuinely help?)* + +  + +  + +  + +  + +  + +--- + +**2. Here is what I think it should not be used for.** + +*(Where does it cause harm? Where does it replace something that shouldn't +be replaced? What decisions should stay with humans?)* + +  + +  + +  + +  + +  + +--- + +**3. Here is what I am willing to do about it.** + +*(This can be small. "I will ask where training data comes from." +"I will tell one person what I learned." "I will read the terms of service." +"I will choose not to use a tool that I know harms a specific community." +What will you actually do?)* + +  + +  + +  + +  + +  + +--- + +*Name (or handle): _______________________* +*Date: _______________________* +*This manifesto represents my thinking as of today.* +*I reserve the right to change my mind when I learn more.* + +--- + +## A Note on Manifestos + +A manifesto is a public declaration of beliefs and intentions. +The word comes from the Latin *manifestus* — made clear, made visible. + +The most famous manifestos in history were written by people who believed +the stakes were high enough to demand clarity — not hedged academic +prose, not balanced summaries, but a clear statement: *here is what I +think, here is why, here is what I intend to do.* + +You are living through a moment when the stakes are genuinely high. +What you write today does not have to change the world. +It has to be honest. + +--- + +## Whole-Series Reflection (Optional, 10 min add-on) + +If time allows after the share-out, ask students to return to the very +first question from Lesson 01: + +> *"It doesn't know anything. It's guessing."* + +*Do you agree now? Disagree? How has your answer changed — or stayed +the same — over the course of the series?* + +This full-circle moment lands differently after six lessons than it did +at the start. Give it space. + +--- + +## Standards Alignment + +| Framework | Standard | +|-----------|----------| +| CSTA K-12 CS Standards | 3A-IC-24 (ethical impacts), 3A-IC-25 (social effects of computing) | +| ISTE Student Standards | 1.2a (digital citizenship), 1.7d (understand AI limitations and implications) | +| Common Core ELA (9-10) | W.9-10.1 (write arguments), SL.9-10.4 (present information) | +| Common Core ELA (11-12) | W.11-12.1 (write arguments to support claims), SL.11-12.1 (collegial discussion) | +| NCSS Social Studies | Civic Ideals and Practices; Power, Authority, and Governance | + +--- + +## Series Complete + +| # | Title | Theme | Status | +|---|-------|-------|--------| +| 01 | The Oracle That Guesses | How LLMs work | ✓ Complete | +| 02 | Whose Voice Is This? | Training data and bias | ✓ Complete | +| 03 | The Consent Ledger | Data, privacy, who benefits | ✓ Complete | +| 04 | The Mirror Test | AI, identity, consciousness | ✓ Complete | +| 05 | The Unfinished Map | Evaluating AI output | ✓ Complete | +| 06 | After the Tool | Agency and manifesto | ✓ Complete | + +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* + +--- + +## Agent Disclosure + +This lesson was drafted with AI assistance (Claude, Anthropic). +Human direction, editorial judgment, and all instructional design decisions +are those of Sean Campbell (`rudi193-cmd`). The closing address in Part 4 +was drafted collaboratively and revised by the human author. +All facilitation guidance and the manifesto prompt are human-authored. + +--- + +## License + +This work is licensed under +[Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). + +Free to share, adapt, and use in any context — including commercially — +with attribution. + +*Part of the AI Literacy for High School series.* +*Series index: [ai-literacy-9-12-index.md](./ai-literacy-9-12-index.md)* diff --git a/lessons/cs-k12-the-scribe-who-forgot-his-dreams.md b/lessons/cs-k12-the-scribe-who-forgot-his-dreams.md index ad5682d..3571349 100644 --- a/lessons/cs-k12-the-scribe-who-forgot-his-dreams.md +++ b/lessons/cs-k12-the-scribe-who-forgot-his-dreams.md @@ -39,6 +39,8 @@ Students will understand — through story, not jargon — that many AI systems **Co-authorship:** Story developed by Sean Campbell with Hanz Christain Anderthon (Professor of Computational Kindness, UTETY). AI tools assisted drafting; **human direction and edit are authoritative.** +**Planning / presentation thread:** [`research/emerging-rule-scribe-talkthrough.md`](../research/emerging-rule-scribe-talkthrough.md) · [`research/emerging-rule-presentation-scribe-lesson.md`](../research/emerging-rule-presentation-scribe-lesson.md) + --- ## The Story @@ -184,6 +186,62 @@ He was sure of that also. --- +## Assessment + +This lesson assesses **understanding through discussion and expression**, not a written test. + +**Evidence you might document:** + +- **Verbal (K–12):** Student explains in their own words that the scribe can be helpful *and* not remember prior visits — without using jargon. +- **Reflective (3–12):** Student names which visitor response they relate to (stopped / came more / adjusted) and gives a reason. +- **Visual (K–5):** Drawing of what stayed vs. what dissolved overnight (knowledge vs. the day's visitors). +- **Optional tech (6–12):** Student describes what felt the same and what was missing after two fresh chat sessions with the same question. + +**Sample report-card language:** + +- *The student articulated understanding after reflection, indicating durable sense-making.* +- *The student demonstrated conceptual transfer by connecting ideas* (e.g., scribe ↔ chat tool they have used). + +Holistic participation is sufficient. There is no single correct moral; the goal is **accurate notice** of how the system behaves. + +--- + +## Differentiation + +### K–2 +- Teacher reads aloud; pause after the woman and the boy. +- Draw: *What did the scribe know? What did he forget?* +- One question: *Was the scribe mean?* (Expected: no — he is made this way.) + +### 3–5 +- Read aloud or paired reading. +- Discuss the three visitor types at the end. +- Write one sentence: *Would you still visit the scribe? Why?* + +### 6–8 +- Students read silently or in pairs. +- Socratic discussion on memory, trust, and privacy. +- Optional: fresh-chat demo (teacher-led first). + +### 9–12 +- Full discussion prompts; connect to tools students already use. +- Extension: *Sequence, not duration* — the gap between turns is not experienced by the system as waiting (see `research/notes-for-future-lessons.md`). +- Optional writing: personal rule for when stateless help is useful vs. when you need a human who remembers context. + +### ELL / multilingual +- Story is read-aloud friendly; allow home-language discussion before English exit ticket if your school policy allows. +- Key concept to translate carefully: *remember the conversation* vs. *know the topic*. + +### Support +- Provide illustrated storyboard of four beats: help → sleep → forget → return. +- Pre-teach: *stateless* is not required; use *does not keep your story overnight*. + +### Extension +- Students write a fourth visitor with a different response to discovering the scribe forgets. +- Compare: when is forgetting a feature (privacy) vs. a problem (medical follow-up)? + +--- + ## License This lesson is shared under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). You may adapt and share with attribution. diff --git a/research/emerging-rule-presentation-scribe-lesson.md b/research/emerging-rule-presentation-scribe-lesson.md new file mode 100644 index 0000000..6ee384e --- /dev/null +++ b/research/emerging-rule-presentation-scribe-lesson.md @@ -0,0 +1,126 @@ +# Lesson Proposal — Emerging Rule Community + +**To:** Emerging Rule team +**From:** Sean Campbell (with UTETY faculty co-authorship) +**Re:** Issue #5 — Computer Science / how AI works (K–12) +**Date:** 2026-05-26 +**License:** CC BY 4.0 + +--- + +## What we're proposing (60 seconds) + +A **read-aloud parable lesson** that teaches *how AI behaves* — warm, knowledgeable, **does not remember yesterday** — with **no devices required** for the core lesson. + +**Title:** [*The Scribe Who Forgot His Dreams*](../lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) +**Closes:** [Issue #5 — CS / coding lesson](https://github.com/Emerging-Rule/community/issues/5) +**Co-author voice:** Hanz Christain Anderthon, Professor of Computational Kindness (UTETY editorial persona + human direction) + +This fills a gap your current library doesn't cover: every shipped lesson uses AI *in* the activity. This one teaches what the system *is* before anyone opens a chatbot. + +--- + +## Why this fits Emerging Rule + +| Your repo today | This lesson | +|-----------------|-------------| +| Story Problem Factory — catch AI math errors | Scribe — understand AI **memory and trust** | +| Who Wrote This? — authorship (9–12) | Scribe — **K–12**, story-first | +| Arguing with AI — fact-check claims | Scribe — **no debate**, documents behavior | +| Requires devices for core activity | **No devices required** (optional 6–12 demo) | + +**Posole criterion:** A teacher with 30 students and no prep period can run this **tomorrow** — read aloud + discussion. + +**Mission alignment:** Students remain the thinkers. The adult remains the adult. The lesson **documents the gap**; it does not sell a product or train prompt engineering. + +--- + +## Lesson at a glance + +| Field | Value | +|-------|--------| +| **Grade** | K–12 (differentiation by band in lesson file) | +| **Subject** | CS / AI literacy / digital citizenship | +| **Duration** | 30–45 min | +| **AI tool** | None required | +| **Format** | Parable → discussion → optional extension | +| **File** | `lessons/cs-k12-the-scribe-who-forgot-his-dreams.md` | + +--- + +## What students leave knowing + +Not definitions. **Noticed behavior:** + +1. Something can **help you completely** in the moment and **not remember you** later. +2. That is a **design property**, not meanness or brokenness. +3. People respond differently once they notice (stop / come more / adjust) — all valid. +4. **Your** memory of the conversation and **its** memory of you are not the same thing. + +Optional (6–12): fresh chat session demo makes the abstract concrete. + +--- + +## 45-minute run of show (presentation flow) + +Use this as the **meeting walkthrough** or pilot agenda. + +| Time | What happens | Your talking point | +|------|----------------|-------------------| +| **0–2 min** | Hook | *"Can something know everything and still not know you?"* | +| **2–18 min** | Read parable aloud | Pause after the woman, the boy, and the forgetting night. | +| **18–28 min** | Discussion | Three prompts at end of story (memory, enough?, what would change?) | +| **28–35 min** | Visitor types | Who stopped / came more / adjusted — no wrong answer. | +| **35–42 min** | Connect | *"Have you used something that felt like the scribe?"* (optional share) | +| **42–45 min** | Close | One sentence exit: *"What will you remember that it won't?"* | + +**Optional extension (6–12, +10 min):** Two fresh chat sessions, same question — compare. + +--- + +## Assessment (how you know it worked) + +No quiz. Observable evidence: + +- Student explains **helpful but doesn't remember** without jargon. +- Student picks a visitor type and defends it. +- (Younger) drawing: what stayed vs. what dissolved. + +Uses **expressive pathways** — verbal, reflective, visual — not output-only grading. Language compatible with assessment-visibility framing (see `research/assessment-visibility-v1.1/` if useful for teacher PD; not student-facing). + +--- + +## What we're asking from Emerging Rule + +1. **Review** the lesson file and this proposal — feedback welcome on tone, grade bands, Issue #5 fit. +2. **Merge** into `/lessons` on `community` (PR ready in contributor worktree). +3. **Credit:** Sean Campbell + Hanz Christain Anderthon (UTETY); CC BY 4.0. +4. **Optional:** One classroom or team pilot; share *Notes from the Classroom* back to the repo. + +We are **not** asking for product placement, LevelShip requirement, or UTETY platform dependency. + +--- + +## What comes next (if this lands) + +| Wave | Item | Issue | +|------|------|-------| +| Ready | AI & Education reading list (teachers) | #4 | +| Planned | MS History — *Three Books, Three Wars* (Oakenscroll) | #3 | +| Planned | Science 3–5 ecosystem fact-check | #1 | +| Blocked | Spanish-only El Jardín — needs classroom Spanish author | #2 | +| Extension | *Sequence, not duration* (Scribe follow-on, 6–12) | — | + +Full roadmap: `research/lesson-roadmap.md` + +--- + +## Appendix — full lesson + +The complete contributor-ready lesson (story + For Teachers + assessment + differentiation): + +**[lessons/cs-k12-the-scribe-who-forgot-his-dreams.md](../lessons/cs-k12-the-scribe-who-forgot-his-dreams.md)** + +--- + +*Human author: Sean Campbell · Editorial persona: Hanz Christain Anderthon · Agent-assisted drafting with human direction and edit authoritative.* diff --git a/research/emerging-rule-scribe-talkthrough.md b/research/emerging-rule-scribe-talkthrough.md new file mode 100644 index 0000000..f6502b7 --- /dev/null +++ b/research/emerging-rule-scribe-talkthrough.md @@ -0,0 +1,155 @@ +# Emerging Rule — Scribe Lesson Talk-Through (saved thread) + +**Status:** Paused — ready to resume when meeting happens +**Saved:** 2026-05-26 +**Related files:** +- Lesson (repo): [`lessons/cs-k12-the-scribe-who-forgot-his-dreams.md`](../lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) +- Presentation brief: [`emerging-rule-presentation-scribe-lesson.md`](emerging-rule-presentation-scribe-lesson.md) +- Roadmap: [`lesson-roadmap.md`](lesson-roadmap.md) +- Git branch with original merge: `feat/scribe-lesson-hanz` (content now in `/lessons` on worktree) + +--- + +## What we're bringing them + +Not “here’s a markdown file.” The pitch: + +> **Issue #5 is open.** The repo teaches kids to *use* AI. Nobody teaches them *what it is* — without jargon, without devices, tomorrow morning. + +**The Scribe** is that lesson. + +--- + +## 60-second pitch (talk track) + +*“Every lesson in the repo assumes a chatbot is already open. This one comes first — or stands alone. It’s a story: a scribe who knows everything and forgets everyone overnight. Warm, helpful, no memory of your day. Kids notice the gap themselves. No LLM slide deck. No coding. Read aloud, discuss, done. Optional demo for older kids: open a fresh chat twice and compare.”* + +**Tone check when resuming:** Does this sound like Sean talking to them? Too literary? Too UTETY? + +--- + +## Why they might care + +| Their gap | Scribe fills it | +|-----------|-----------------| +| CS / how AI works (#5) | Empty slot | +| Device access uneven | Core lesson needs **zero** tech | +| Youngest grades | K–12, not only 9–12 SEL | +| “AI literacy” buzzword fatigue | Story, not vocabulary | + +**Expected pushback:** *“It’s not really CS.”* +**Counter:** Issue #5 says “CS **or coding**” — this is **systems behavior** literacy (AI4K12 Big Ideas territory without the poster). + +--- + +## 45-minute run of show (pilot or live demo for their team) + +| Time | What happens | Talking point | +|------|----------------|---------------| +| 0–2 min | Hook | *“Can something know everything and still not know you?”* | +| 2–18 min | Read parable aloud | Pause after the woman, the boy, the forgetting night | +| 18–28 min | Discussion | Three questions at end of story | +| 28–35 min | Visitor types | stopped / came more / adjusted — no wrong answer | +| 35–42 min | Connect | *“Have you used something that felt like the scribe?”* | +| 42–45 min | Close | *“What will you remember that it won’t?”* | + +**Optional +10 min (6–12):** Two fresh chat sessions, same question — compare. + +**Open when resuming:** Live read for their team vs. document they approve only? + +--- + +## What students leave knowing (keep repeating) + +Not “stateless inference.” Four notices: + +1. Helpful **now** ≠ remembers **later** +2. That’s **how it’s made**, not cruelty +3. People react differently — all valid +4. **Your** continuity ≠ **its** continuity + +--- + +## Hanz / attribution (decide at meeting time) + +Co-author: **Hanz Christain Anderthon** — parable teacher voice; Sean = human author of record. + +| Level | What to say | +|-------|-------------| +| **Minimal** (recommended lead) | “Co-developed with an editorial persona for kid-safe parable voice; I’m the human author of record.” | +| **Medium** | “UTETY faculty voice for fables — same production model as the reading list annotations.” | +| **Full** | UTETY / LevelShip / Chitlins — only if they already know that world | + +**Default:** Lead with Sean + CC BY 4.0; Hanz as *voice*, not franchise. + +--- + +## The ask (keep small) + +1. **Does this close #5?** (wording / subject tag) +2. **Merge to `/lessons`** with attribution +3. **Optional:** One team member runs it once → *Notes from the Classroom* + +**Not asking:** LevelShip install, platform tie-in, exclusive partnership. + +--- + +## Decisions deferred (pick up when thread resumes) + +- [ ] **Who is in the room?** Founders / community managers / teacher advisors → sets pitch vs. pedagogy balance +- [ ] **Library ordering:** Scribe as *first* “how AI works” lesson vs. one catalog entry among many +- [ ] **Assessment framing:** Verbal/reflective/visual in presentation, or teachers-only in lesson file +- [ ] **Sequence-not-duration:** Same meeting as advanced extension, or Scribe v2 (`research/notes-for-future-lessons.md`) +- [ ] **Story beats:** Trim length? Woman / war books — any school sensitivity to flag +- [ ] **Wave 2 in same meeting?** Reading list (#4) + roadmap, or **Scribe only** until they bite +- [ ] **15-min vs 45-min** presentation slot — tighten brief if needed +- [ ] **PR / merge:** Open PR from worktree when approved (`feat/research-ai-education-reading-list` has local files; scribe also on `feat/scribe-lesson-hanz`) + +--- + +## Repo state when saved + +| Item | Location | Notes | +|------|----------|-------| +| Lesson file | `lessons/cs-k12-the-scribe-who-forgot-his-dreams.md` | Merged from branch + Assessment + Differentiation | +| Presentation brief | `research/emerging-rule-presentation-scribe-lesson.md` | Meeting-oriented | +| README | Updated — 6 lessons, Scribe in table | Not pushed | +| Reading list | `research/ai-education-reading-list.md` | Separate PR (#4) | +| KB atoms | `432EF091` (Scribe), `0F2B4BEA` (Oakenscroll plan) | Willow KB | + +--- + +## Source + +Imported from `~/Desktop/Nest/files.zip` (2026-05-27). Full 6-lesson HS arc + index. + +| # | Lesson | File | +|---|--------|------| +| — | **Series index** | [`lessons/ai-literacy-9-12-index.md`](../lessons/ai-literacy-9-12-index.md) | +| 01 | The Oracle That Guesses | `ai-literacy-hs-01-the-oracle-that-guesses.md` | +| 02 | Whose Voice Is This? | `ai-literacy-hs-02-whose-voice-is-this.md` | +| 03 | The Consent Ledger | `ai-literacy-hs-03-the-consent-ledger.md` | +| 04 | The Mirror Test (+ Scribe pre-read) | `ai-literacy-hs-04-the-mirror-test.md` | +| 05 | The Unfinished Map | `ai-literacy-hs-05-the-unfinished-map.md` | +| 06 | After the Tool (manifesto capstone) | `ai-literacy-hs-06-after-the-tool.md` | + +**Arc:** mechanics → ethics → agency. Posole criterion throughout. Issue #5 at K–12 via Scribe; this series **supersedes** “HS-only CS gap” for grades 9–12. + +**Presentation note:** When meeting with Emerging Rule, lead with **Scribe** (all grades, zero devices) → **this series** (9–12 depth). See talkthrough below. + +--- + +## Resume prompt + +When the meeting is scheduled, start with: + +1. Re-read this file + presentation brief +2. Confirm audience and time slot +3. Choose attribution level (Hanz) +4. Run or summarize the 45-min flow +5. Close with the three asks +6. If yes → PR + comment on Issue #5 + +--- + +*Saved from Cursor talk-through with Sean Campbell · 2026-05-26* diff --git a/showcases/ai-literacy-9-12/README.md b/showcases/ai-literacy-9-12/README.md new file mode 100644 index 0000000..f539881 --- /dev/null +++ b/showcases/ai-literacy-9-12/README.md @@ -0,0 +1,64 @@ +# Example Submission — AI Literacy for High School (9–12) + +**Status:** Example · submitted for community review · not yet classroom-piloted through Emerging Rule +**Contributor:** Sean Campbell (`rudi193-cmd`) +**License:** CC BY 4.0 +**Related issue:** [#5 — CS / how AI works](https://github.com/Emerging-Rule/community/issues/5) + +--- + +## What this is + +A **six-lesson example arc** for grades 9–12, plus a **K–12 companion parable** (*The Scribe Who Forgot His Dreams*). Posted as an **example submission** so educators and maintainers can review structure, tone, and fit before any “official” merge decision. + +Each lesson meets the **posole criterion**: usable by a teacher with 30 students, no devices required for the core activity. + +--- + +## Files in this showcase + +| Item | Location | +|------|----------| +| **Series index** | [`lessons/ai-literacy-9-12-index.md`](../lessons/ai-literacy-9-12-index.md) | +| Lesson 01 — The Oracle That Guesses | [`lessons/ai-literacy-hs-01-the-oracle-that-guesses.md`](../lessons/ai-literacy-hs-01-the-oracle-that-guesses.md) | +| Lesson 02 — Whose Voice Is This? | [`lessons/ai-literacy-hs-02-whose-voice-is-this.md`](../lessons/ai-literacy-hs-02-whose-voice-is-this.md) | +| Lesson 03 — The Consent Ledger | [`lessons/ai-literacy-hs-03-the-consent-ledger.md`](../lessons/ai-literacy-hs-03-the-consent-ledger.md) | +| Lesson 04 — The Mirror Test | [`lessons/ai-literacy-hs-04-the-mirror-test.md`](../lessons/ai-literacy-hs-04-the-mirror-test.md) | +| Lesson 05 — The Unfinished Map | [`lessons/ai-literacy-hs-05-the-unfinished-map.md`](../lessons/ai-literacy-hs-05-the-unfinished-map.md) | +| Lesson 06 — After the Tool | [`lessons/ai-literacy-hs-06-after-the-tool.md`](../lessons/ai-literacy-hs-06-after-the-tool.md) | +| **K–12 companion** | [`lessons/cs-k12-the-scribe-who-forgot-his-dreams.md`](../lessons/cs-k12-the-scribe-who-forgot-his-dreams.md) | + +**Presentation brief (maintainers):** [`research/emerging-rule-presentation-scribe-lesson.md`](../research/emerging-rule-presentation-scribe-lesson.md) + +--- + +## Arc (one line per unit) + +1. **Mechanics** — prediction ≠ knowledge +2. **Representation** — whose voices trained the model +3. **Data** — consent and who benefits +4. **Philosophy** — simulation vs understanding (pairs with Scribe) +5. **Practice** — evaluate confident wrong answers +6. **Agency** — student manifestos + +--- + +## How to give feedback + +- Comment on the pull request +- Open an issue tagged `lesson` +- Email admin@emergingrule.com with subject `[Example Feedback] AI Literacy HS` + +--- + +## What we're asking + +This is an **example post**, not a claim of classroom validation. We welcome: + +- Fit with repo standards (template, prompts, bilingual goals) +- Whether to list in main README lesson table vs. showcase-only until piloted +- Spanish priority (index notes L01 + L04 for bilingual release) + +--- + +*Human direction and editorial judgment: Sean Campbell. AI-assisted drafting and formatting disclosed in individual lesson files.*