Course: AI Fluency for Educators

Mirrors: Anthropic Academy — AI Fluency for Educators · https://anthropic.skilljar.com/ Audience: Teachers, instructors, and instructional designers at any level. No technical background needed. · Time: ~60 min + project Prereqs: claude-101 and a claude.ai account (setup-checklist). · Backing notes: ai-fluency-framework-foundations, ai-capabilities-and-limitations Project: p03-ai-fluency-delegation-audit

This course applies the 4D framework — Delegation, Description, Discernment, Diligence (an open educational resource by Rick Dakan & Joseph Feller) — to the work of teaching. The throughline: educators play a double game. You use AI to do your own work better, and you model fluent, honest AI use for the students watching you. Both matter.

Learning objectives

After this course you can:

• Apply the four Ds to lesson design, materials, and assessment.
• Delegate prep work to AI while keeping pedagogical judgment yourself.
• Redesign assignments so they remain meaningful in an AI-saturated world.
• Build AI-resilient assessments that measure understanding, not output.
• Model good Delegation and Diligence so students learn integrity by watching you.
• Weigh equity and access so AI narrows rather than widens gaps.

---

Module 1 — The 4D framework, for teaching

🎞 Frame 1 · Delegation: what teaching work is AI's, and what's yours · ⏱ ~3 min

🎬 Scene — A teacher hands AI a stack of routine prep — a quiz draft, three reading-level variants, a rubric skeleton — and keeps the judgment about what students need.

🧠 Concept — Delegation is dividing labor between you and AI. AI is strong at first drafts, variants, and reformatting; you own learning goals, knowing your students, and the call on what's good enough. Delegate the production, keep the pedagogy.

🖼 On screen

Delegate to AI	Keep as the educator
Draft quizzes, worked examples, rubric scaffolds	Choosing learning objectives
Reading-level and language variants	Knowing this class & these students
Reformatting, summarizing standards	Final judgment on quality & fairness
Brainstorming activity ideas	The relationship and the feedback that lands

✅ Checkpoint — Name one prep task you'd delegate this week and one pedagogical decision you'd never delegate.

🎞 Frame 2 · Description: prompt with your pedagogy · ⏱ ~3 min

🎬 Scene — "Make a worksheet on fractions" yields generic filler. A described prompt — grade, standard, misconceptions to target, scaffolding — yields something usable.

🧠 Concept — Description means giving AI your context: grade level, standard or objective, the misconceptions you want to address, prior knowledge, and the output format. The more of your teaching expertise you encode, the more the output reflects it. (Deeper: prompt-engineering-basics.)

🖼 On screen

❌ "Make a worksheet on fractions."
✅ "Create a 6-question worksheet for 5th graders on adding
    fractions with unlike denominators. Target the common
    mistake of adding denominators. Scaffold from concrete
    (visual) to abstract. Include an answer key with the
    misconception each wrong answer reveals."

✅ Checkpoint — Take a lesson you teach and list the three context facts AI most needs to draft good materials for it.

🎞 Frame 3 · Discernment & Diligence: you are the editor of record · ⏱ ~3 min

🎬 Scene — AI produces a clean-looking reading passage with a subtle factual error and a too-hard vocabulary word. The teacher catches both.

🧠 Concept — Discernment is reviewing AI output for accuracy, bias, and fit before it reaches students. Diligence is the responsibility around it: fact-checking, watching for bias, citing sources, and being transparent with students and colleagues about AI's role. Nothing goes to students unvetted. (Deeper: ai-capabilities-and-limitations.)

🖼 On screen

Before it reaches a student, check:
  Discernment → accurate? age-appropriate? on-objective? unbiased?
  Diligence   → sources verified? AI role disclosed where it matters?

⚠️ Gotcha — Polished prose hides errors. The smoother the draft, the more deliberately you must check it.

✅ Checkpoint — Why is the educator always the "editor of record" for AI-generated materials?

---

Module 2 — Lesson and materials workflows

🎞 Frame 4 · Differentiation at speed · ⏱ ~3 min

🎬 Scene — One core reading becomes three versions — below, at, and above grade level — plus a home-language summary, in minutes.

🧠 Concept — AI's highest-value teaching use is differentiation. Producing tiered materials by hand is the work teachers never have time for; delegating the variants (while you set the targets) makes inclusive teaching feasible.

🖼 On screen

"Take this passage. Produce three versions: (a) two grade levels
 below, (b) on level, (c) an extension for advanced readers.
 Keep the same key vocabulary in all three. Flag any term that
 may need pre-teaching."

✅ Checkpoint — Which of your materials would benefit most from a two-minute tiered rewrite?

🎞 Frame 5 · From standard to scaffolded lesson · ⏱ ~3 min

🎬 Scene — A teacher pastes a standard and gets a lesson arc — hook, model, guided practice, independent practice, check — that they then edit.

🧠 Concept — AI is a fast co-planner. Give it the standard and your constraints (time, class size, materials), get a structured draft, then apply your judgment. The draft saves the blank-page time; the editing is where your expertise lives.

🖼 On screen

"Draft a 45-minute lesson for [standard]. Structure: hook,
 I-do, we-do, you-do, exit ticket. Note where students
 typically struggle and suggest a quick check at each phase.
 I'll edit — keep it concrete, not generic."

⚠️ Gotcha — A generated lesson is a starting point, never a script to read cold. Always pass it through your own filter.

✅ Checkpoint — Name the part of a generated lesson you'd most expect to rewrite, and why.

🎞 Frame 6 · Feedback and rubric workflows · ⏱ ~3 min

🎬 Scene — A teacher uses AI to draft rubric-aligned feedback comments, then personalizes each before sending.

🧠 Concept — AI can accelerate feedback by drafting rubric-aligned comments and spotting patterns across a set of responses — but the relationship and the final judgment stay yours. Use it to do more feedback, not to remove yourself from it.

🖼 On screen

"Here's my rubric and a student response. Draft feedback that
 names one strength and one specific next step tied to the
 rubric. Neutral, encouraging tone. I'll personalize it."

✅ Checkpoint — What must a teacher add to AI-drafted feedback before a student sees it?

---

Module 3 — AI-resilient assessment & assignment redesign

🎞 Frame 7 · The shift: assess the process, not just the product · ⏱ ~3 min

🎬 Scene — An essay-only assignment (easy to outsource to AI) becomes essay + in-class defense + annotated drafts.

🧠 Concept — If an assignment can be completed by pasting the prompt into an AI, it no longer measures what you think. Redesign toward process evidence: drafts, oral defenses, in-class application, personal reflection, and work tied to your specific class.

🖼 On screen

AI-fragile	AI-resilient
Take-home factual essay	Essay + 3-minute oral defense
"Summarize this article"	"Apply this to our local case study"
One final product	Drafts + reflection on revisions
Generic prompt	Prompt tied to in-class discussion

✅ Checkpoint — Take one current assignment and name one process-evidence element you could add.

🎞 Frame 8 · Designing assignments that use AI well · ⏱ ~3 min

🎬 Scene — Instead of banning AI, an assignment requires students to critique an AI's answer and document where it was wrong.

🧠 Concept — AI-resilience isn't only about prevention. The strongest assignments make AI part of the task: have students prompt it, discern its errors, and reflect — turning the assessment into a Discernment exercise. This teaches fluency while staying rigorous.

🖼 On screen

Assignment idea: "Ask AI to solve this problem. It will make at
 least one error. Find the error, explain why it's wrong, and
 produce the correct solution. Submit your prompt, the AI's
 answer, and your correction."

✅ Checkpoint — Sketch a one-sentence assignment that requires students to correct an AI rather than copy it.

🎞 Frame 9 · Model the Diligence you want to see · ⏱ ~3 min

🎬 Scene — A teacher shows the class the AI draft of a handout and the changes they made and why, narrating their Discernment out loud.

🧠 Concept — Students learn integrity from what you do, not what you forbid. When you disclose your own AI use, show your verification, and name what you kept human, you teach the 4Ds by example. Transparency from the front of the room normalizes it.

🖼 On screen

"I used AI to draft this. Here's what it got wrong, here's what
 I changed, and here's the part I wrote myself because it
 needed my judgment."

✅ Checkpoint — Describe one moment in your teaching where you could narrate your own Discernment for students.

---

Module 4 — Equity, access, and responsibility

🎞 Frame 10 · The access gap is an equity issue · ⏱ ~3 min

🎬 Scene — Two students: one with a paid AI plan and a quiet study space, one with neither. The same "use AI" assignment lands unequally.

🧠 Concept — Diligence for educators includes equity. Access to capable AI, devices, connectivity, and even the literacy to prompt well is unevenly distributed. Design so AI use is supported and optional-equivalent, not an assumed resource that penalizes those without it.

🖼 On screen

Equity checklist:
  • Does this assignment require paid AI? (avoid)
  • Is there a no-AI path to the same learning?
  • Do I teach prompting, not assume it?
  • Are data/privacy expectations clear for minors?

✅ Checkpoint — Name one way an AI assignment could unintentionally disadvantage some students.

🎞 Frame 11 · Student data & privacy · ⏱ ~2 min

🎬 Scene — A teacher about to paste a class roster with names into a chat — and stops.

🧠 Concept — Don't put identifiable student data (names, grades, records, anything protected by privacy law like FERPA) into AI tools without institutional approval. De-identify before you delegate. This is non-negotiable Diligence when minors and protected records are involved.

⚠️ Gotcha — "It's just for feedback" is not a waiver. Strip identifying details first.

✅ Checkpoint — What should you remove before pasting a student's work into an AI tool?

🎞 Frame 12 · You've got the educator's 4Ds · ⏱ ~1 min

🎬 Scene — Recap slide: delegate prep, describe your pedagogy, discern every output, do the diligence — and model it.

🧠 Concept — You can now use AI to teach more inclusively and assess more meaningfully, while modeling the integrity you want from students. Next: audit it in the project.

✅ Checkpoint — Without looking, list the four Ds and one teaching example of each.

---

🛠 Project

Complete p03-ai-fluency-delegation-audit — Delegation & Discernment Audit (Educator edition). You'll take one real assignment or lesson, map what you delegate to AI vs. what you keep, redesign one AI-fragile assessment into an AI-resilient one, vet a generated material for accuracy/bias/equity, and write a transparency note modeling your own AI use for students.

🧪 Self-check quiz

1. What are the four Ds, in order?
2. What teaching work is safe to delegate, and what must stay with the educator?
3. What makes an assignment "AI-fragile," and one way to make it resilient?
4. Why is the educator always the "editor of record" for AI output?
5. Give one way to design an assignment that uses AI rather than banning it.
6. Name two equity risks when assigning AI-based work.
7. What must you do before pasting student work into an AI tool?

<details><summary>Answers</summary>

1. Delegation, Description, Discernment, Diligence. 2. Delegate drafts/variants/reformatting; keep objectives, knowing your students, and final quality/fairness judgment. 3. AI-fragile = completable by pasting the prompt into AI; add process evidence (drafts, oral defense, in-class application). 4. AI output can carry errors, bias, and poor fit; nothing reaches students unvetted. 5. Have students critique/correct an AI answer and document the error. 6. Unequal access to paid AI/devices/connectivity; differing prompting literacy (also: data/privacy for minors). 7. De-identify it — remove names and protected details (FERPA/privacy).

</details>

🎓 Certificate criteria

You've "passed" AI Fluency for Educators when you can:

• State the four Ds with a teaching example of each.
• Generate a differentiated material and vet it for accuracy, bias, and fit.
• Redesign one AI-fragile assignment into an AI-resilient one.
• Model your own AI use transparently for students.
• Complete p03-ai-fluency-delegation-audit and journal one assessment you redesigned in learning-journal-template.

Tick this course off in progress and record the date you earned Anthropic's official certificate.

🔗 Sources & deeper notes

• Official course: https://anthropic.skilljar.com/ (AI Fluency for Educators)
• Framework: open educational resource by Rick Dakan & Joseph Feller
• Vault notes: ai-fluency-framework-foundations, ai-capabilities-and-limitations, prompt-engineering-basics
• Project: p03-ai-fluency-delegation-audit · Related courses: ai-fluency-for-students, teaching-ai-fluency