AI Fluency for your world
How the 4D framework applies to builders, educators, students, nonprofits and more.
Series Claude Academy: Become AI fluent Part 4 of 4
A walkthrough of the Claude Academy course AI Fluency for your world (Become AI fluent track, 9 audience-specific courses). Each lesson is summarised with my own tables and examples; take the course for the full lessons. All the courses are listed in the Claude Academy course index.
Note: Detailed content expanded from course outline; verify against lessons. (Public course pages and lesson summaries/key takeaways were the primary source; comparison tables and practice ideas are added study aids.)
Sources:
- Builders: https://academy.claude.com/courses/ai-fluency-for-builders
- Creative work: https://academy.claude.com/courses/ai-fluency-for-creative-work
- Educators (higher ed): https://academy.claude.com/courses/ai-fluency-for-educators
- pK–12 educators: https://academy.claude.com/courses/ai-fluency-for-k-12-educators
- pK–12 Train the Trainer: https://academy.claude.com/courses/ai-fluency-for-pk-12-train-the-trainer
- Teaching AI Fluency: https://academy.claude.com/courses/teaching-ai-fluency
- Nonprofits: https://academy.claude.com/courses/ai-fluency-for-nonprofits
- Small businesses: https://academy.claude.com/courses/ai-fluency-for-small-businesses
- Students: https://academy.claude.com/courses/ai-fluency-for-students
Audience: Learners who want the AI Fluency Framework applied to a specific role. Each course covers the foundations “lightly” and recommends AI Fluency: Framework & Foundations first. This page focuses on what is specific to each audience.
Shared core (brief refresher)
| D | Sub-components | One-liner |
|---|---|---|
| Delegation | Problem Awareness · Platform Awareness · Task Delegation | Decide what you do, what AI does, and what you do together |
| Description | Product · Process · Performance | Tell the AI what you want, how to approach it, and how to behave |
| Discernment | Product · Process · Performance | Judge the output, the reasoning, and the AI’s behavior |
| Diligence | Creation · Transparency · Deployment | Choose tools responsibly, disclose AI’s role, own what you ship or share |
- Interaction modes: Automation (AI executes specific tasks), Augmentation (you and AI think and create together), Agency (you configure AI’s knowledge and behavior so it works independently on your behalf).
- Two nested loops (used across most variants):
- Outer loop — Delegation ↔ Diligence: strategic and ethical decisions about whether, when, and how to use AI, and owning the result.
- Inner loop — Description ↔ Discernment: the moment-to-moment craft of describing, evaluating, and refining.
- Recurring themes: build a reusable context document, stay “the human in the loop,” and treat the 4Ds as “a loop, not a line.”
At-a-glance comparison
| Course | Partner(s) | Distinctive tools/artifacts | Signature scenario |
|---|---|---|---|
| Builders | CodePath | Builder brief, acceptance tests, Description Chain, five Discernment lenses, ship/fix/stop | Clinic Wait Time app |
| Creative work | Dakan & Feller | Creative Value Lens, Production Lens, five AI roles, personal AI use policy | An illustrator’s workflows |
| Educators (higher ed) | Dakan & Feller | Teaching context document, AI-assisted course design and materials | Course design → materials → assessments |
| pK–12 educators | Teach For America | Context document, next-token simulator, FERPA/data handling, AI value document | Ms. Okafor, 7th-grade science |
| pK–12 Train the Trainer | AFT | Deck + facilitator guide + learner workbook (grade-band exercises, delegation matrix) | Running a teacher workshop |
| Teaching AI Fluency | Dakan & Feller | Four teaching approaches, outcome/process/reflection assessment, discipline position paper | Teaching the 4Ds to students |
| Nonprofits | GivingTuesday | Data hygiene, validating analysis on known data, organizational AI policy | Maria, James, Rio, Emily |
| Small businesses | PayPal, Prospect Butcher Co., MAKS TIPM Rebuilders | AI-augmented workflow, AI use policy, customer transparency | Brooklyn butcher; Mak (auto power-module rebuilder) |
| Students | Dakan & Feller | AI learning partner, living learning journal, personal AI collaboration policy | Learning and career planning |
1. AI Fluency for Builders
Who: Product people, engineers, designers, and makers. Partner: CodePath (nonprofit focused on building an “AI-native generation of builders”). Modules: Intro & 4Ds → How AI works → 4D framework for builders (Delegation, Description) → Discernment for builders (code, UX) → Tying it together (Diligence) → Wrapping up.
What’s specific:
- Builders own the full arc from customer problem to shipped solution.
- Builder brief: a reusable context document that gives the AI your values, constraints, and context (“onboarding your AI collaborator”).
- Key rule: delegating implementation is usually safe; delegating judgment usually isn’t. Most AI failures trace back to earlier mistakes in description, discernment, or diligence.
- Inner vs. outer loop framing is introduced early.
- Testing the edges: run versatility, hallucination (do the suggested libraries or APIs exist, and are versions current?), and knowledge-cutoff checks in a coding domain you know well.
- Delegation & the builder’s toolkit: decompose the problem first, then decide what role AI plays at each stage. Pick a mode (Automation, Augmentation, Agency) for each toolkit capability. As AI speeds up implementation, your value shifts to framing problems and raising the bar.
- Acceptance tests before code: concrete pass/fail statements (“a patient with a basic smartphone can find the wait time in < 30 s without an account”, not “easy to use”).
- The Description Chain: user voice → requirement → technical spec → AI instruction → tests. Prompting is only one link. “Code works but the product doesn’t” points to a broken link upstream. Tests are the most precise form of description.
- Five Discernment lenses: correctness, quality, fit, experience, responsibility. AI has predictable blind spots in concurrency, security, and anything that only breaks at scale. “It works” isn’t the bar; taste is what sets your work apart.
- UX Discernment: “Make it look good” is a wish, not a spec. AI doesn’t get accessibility right by default, so specify it and audit for it. Check clarity and mental model, information hierarchy, accessibility, and feedback and error handling.
- Diligence = full ownership: “AI wrote it” explains nothing and excuses nothing. Ask about shipping realities (migrations, versioning, rate limits, feature flags). Prototype freely, ship selectively, and be willing to ship it, fix it, or stop it. Tests are the safety net after launch.
- Running project: Clinic Wait Time app: problem brief → delegation plan → acceptance tests → spec and build → user test → ship to 3 real users and iterate.
2. AI Fluency for Creative Work
Who: Writers, illustrators, designers, musicians, and other makers. Authors: Anthropic with Dakan & Feller. Starting stance: begin with no AI, and let analysis decide whether and how to opt in. “No AI” and “AI integration” are both legitimate outcomes. The important decisions happen before the prompt.
What’s specific:
- Creative Value Lens (what the work is for):
| Dimension | Meaning |
|---|---|
| Creative Expression | The experience the work produces |
| Human Connection | The relationship the work facilitates (most tied to authenticity) |
| Distinctiveness | What makes it unlike any other work |
- Authenticity has two parts: the decisions in the work itself and what the audience perceives. Name any mismatch between what you value and what your audience or market values.
- Production Lens (what the work costs): Production Constraints (Time, Effort, Resources) + Technological Complexity (how technology acts on those constraints).
- Technology rearranges constraints rather than simply removing them. AI shifts effort from execution to direction.
- As production constraints fall, the Value Lens matters more. When output is cheap, what sets work apart is its purpose and specificity.
- Inner loop applied to your practice: describe what you make, how you make it, and how you work (Product, Process, Performance), then turn those descriptions into explicit standards. “AI cannot serve a practice you have not described.” Process standards apply at the step level. Performance (your habits as a working creator) is the layer AI affects most over time.
- Outer loop: Problem Awareness first (“what problem am I actually solving?”), then Platform Awareness, then Task Delegation. Apply Creation, Transparency, and Deployment Diligence.
- Pitfalls: convergence (work becoming generic) and skill atrophy in the long term. In the short term, the pitfall is reaching for AI when you don’t know what you want.
- Five AI roles (Delegation outcomes), split by whether AI-generated material appears in the final work the audience encounters:
| No AI material in final work | AI material in final work |
|---|---|
| No AI · Consultant · Tool Builder | Co-Creator · Medium |
Each role has its own Diligence profile.
- Capstone: a personal AI use policy answering where AI fits and doesn’t, what you disclose and to whom, and when you’ll revisit. Decisions are made per mode and per task. “A policy without revision triggers goes stale silently.”
3. AI Fluency for Educators (higher education)
Who: Faculty, instructional designers, and educational leaders with teaching or curriculum responsibilities. Authors: Anthropic with Dakan & Feller. Driving question: How can I apply the framework to my own teaching practice? Modules: Intro & framework review → Course design and learning outcomes → Learning materials and assignments.
What’s specific:
- Why now: students already use AI, employers expect AI fluency, and educators have a timely chance to model thoughtful engagement.
- Emphasis on augmentation over automation, with pedagogical values at the center.
- Teaching context document: a reusable record of your pedagogical values, constraints, and student profile. It turns a generic AI into a teaching and thinking partner.
- 4Ds applied to course design (three tasks): identify essential content and concepts → map the learning journey → write measurable learning objectives.
- Delegation: know what and why before deciding how. Description: invite the AI into your problem space with rich context. Discernment: explain why a suggestion does or doesn’t work for your students. Diligence: keep responsibility and document the collaboration.
- Materials and assignments: lecture slides, study guides, in-class exercises, and assessments built on your established context so they work together coherently. Different materials need different delegation strategies.
- Diligence for educators: protect sensitive data, verify accuracy, check for bias, be transparent about AI’s role, and design assessments that encourage authentic learning when students also have AI.
- The goal is quality and coherence, not just saved time.
4. AI Fluency for pK–12 Educators
Who: pK–12 teachers and school staff. Partner: Teach For America (contributed classroom examples and piloted exercises with teachers). Also recommends AI Capabilities and Limitations. Modules: Intro & 4Ds (incl. “Explore!”) → How AI works → Using AI responsibly → Putting it together → Closure.
What’s specific:
- Reusable context document covering your role, students, and goals. Your pedagogical expertise is your greatest asset.
- How AI works includes an interactive next-token simulator, plus hallucinations and knowledge cutoffs.
- Inner loop example: Ms. Okafor, 7th-grade science, drafts a differentiated informational text for a wide range of readers, then checks it for accuracy, reading level, and alignment to objectives. Upload past materials, give specific context (e.g., “students read below grade level; define key vocabulary in context”).
- Outer loop and ethics: student data protections like FERPA apply to AI tools, so use only district-approved tools for anything containing student information. You’ll face gray areas that policy doesn’t cover. Write personal ethical commitments, and revisit them as students, context, and tools change.
- “AI’s constitution, pedagogy, and you”: different models have different personalities and suit different tasks. Write a short, honest AI value document with your intentions, principles, and boundaries. “AI is not the pedagogy — you are.”
- Tying it together: an end-to-end 4D workflow on a real task. The skill is owning the process, not just the output, and the 4Ds become invisible when used well.
- Closure: use the use case library of blueprints for common educator tasks and pick one real task for this week.
5. AI Fluency for pK–12: Train the Trainer
Who: pK–12 educators preparing to facilitate AI fluency workshops for colleagues. Partner: American Federation of Teachers (AFT), “built by educators, for educators.” Prerequisite: complete AI Fluency for pK–12 Educators first. Modules: Welcome (from learner to trainer) → Train the Trainer Kit (how to use it, deck and facilitator guide, learner workbook).
What’s specific:
- Why: AI decisions made today will shape classrooms for years, so educator voice belongs at the center of those decisions. Educators are closest to the impact, and no algorithm can replicate the relational, human work of teaching.
- The kit:
| Piece | For whom | Use |
|---|---|---|
| Slide deck | Facilitator | The full workshop, ready to project. Includes a personalizable “Meet your facilitator” slide and section navigation |
| Facilitator guide | Facilitator | Key points and activity setup for each slide. Appears as speaker notes and can be downloaded as PDF/HTML |
| Learner workbook | Participants | Warm-up reflection, core 4D activities, a delegation matrix, grade-band exercises, and a closing commitment. Fillable HTML or printable PDF; responses stay on the learner’s device |
- Workshop arc: from opening an honest conversation about AI to committing to a real task.
- Facilitators lead in their own schools and districts, in their own voice and on their own terms. Try the workbook yourself before running the session.
6. Teaching AI Fluency
Who: Educators (faculty, instructional designers, leaders) who already know the framework and want to teach it to students and build capacity across their department or institution. Authors: Anthropic with Dakan & Feller. Modules: Approaches to teaching the framework → Assessing AI Fluency → AI’s impact on disciplinary content.
What’s specific:
- The framework is both descriptive (what happens when people work with AI) and normative (how to do it better). Students need to understand both.
- Four teaching approaches:
| Approach | How it works | Best for |
|---|---|---|
| Linear | Delegation → Description → Discernment → Diligence, step by step | Beginners who need structure |
| Non-linear | Start anywhere and move flexibly between the Ds | More experienced students; reflects real-world complexity |
| Focused | Go deep on a single competency | Depth within limited time |
| Two-loops (loop-based) | Teach the framework as nested strategic and tactical processes | Experienced students |
- Teaching the Delegation-Diligence loop: shift from “how do I use AI?” to “how do I make good decisions about AI?” The loop runs both ways. Teach it through scenarios rather than checklists, and treat constraints as creative catalysts.
- Teaching the Description-Discernment loop: move from commands to conversations that build a cognitive environment (shared vocabulary, established interaction patterns, building on earlier exchanges). Assign multi-interaction work, share your own process, and have students document how their interactions evolved. The strategic loop creates the container that the tactical loop fills.
- Assessing the 4Ds with three complementary approaches:
| Approach | Evaluates | Example assignments |
|---|---|---|
| Outcome-based | What students produce | Improving AI outputs; comparing AI systems |
| Process-based | How students work with AI over time | Annotated chat logs; recorded narrations |
| Reflection-based | Metacognitive awareness | Learning journals; personal policy statements (vary them to avoid fatigue) |
- Assignment design principles: authenticity, iteration, pedagogical transparency (be clear that you assess the collaboration, not only the output). Manage the extra volume with detailed rubrics, peer review, lightning-round conferences, and selective sampling.
- Disciplinary impact: AI affects curriculum, pedagogy, and assessment differently in each field. Ask what gets automated, where partnership adds value, and how students stay accountable. Write a position paper on AI in your discipline.
- Making tacit expertise explicit with colleagues: quality criteria (Discernment), communication norms (Description), how work is broken down (Delegation), ethical standards (Diligence). The goal is to prepare students to be irreplaceable.
7. AI Fluency for Nonprofits
Who: Nonprofit staff in fundraising, communications, programs, operations, and leadership. Partner: GivingTuesday. Context: limited resources, accountability to many stakeholders, mission-driven work. Modules: Intro & 4Ds → Description-Discernment loop (research, writing) → Delegation-Diligence loop (privacy, data analysis) → Putting it together (workflow, integration) → Next steps.
What’s specific:
- Mission-centered: every efficiency gain should translate into greater impact for the communities you serve.
- Researching with AI (Maria, executive director expanding a housing nonprofit from Portland to Seattle): research policy landscapes, funding, and compliance with context-rich prompts. Flag claims to verify and question how recent the information is.
- Writing with AI (James, environmental justice organization, 48-hour emergency grant): upload past successful work to capture the organization’s voice, add details AI can’t know (track record, partnerships, community relationships), and make revisions specific (“add these details about our community relationships”, not “make it more compelling”).
- Privacy and data: some tools train on inputs, and free and paid plans differ in data retention. Match the tool to the sensitivity of the task, break tasks into parts so you can avoid sharing PII, and work backward from the goal to the minimum data needed. If something goes wrong, delete the conversation, request data deletion, and follow your organization’s protocols. Claude’s policies are at privacy.anthropic.com and trust.anthropic.com.
- Data analysis (Rio, Valley Veterans Services): test AI against past data where you already know the answers before trusting it with new questions. Build validated, documented approaches rather than blind trust.
- Workflow automation (Emily, gala email responses): start with Problem Awareness (what are people asking?). Ask “should AI do this?”, not only “can it?”. Documented FAQs can be automated; complaints and high-stakes requests stay with humans. Test with real emails and practice all three types of Diligence.
- Integration: being the human in the loop means making sure AI serves the mission. Avoid dependency through understanding (“can we explain what the AI is doing?”). AI should free time for more human work (handwritten cards, personal outreach). Set norms about what happens to time saved.
- Organizational AI policy covering platforms, task delegation, quality oversight, transparency, and alignment with values. Use blueprints for common nonprofit tasks.
8. AI Fluency for Small Businesses
Who: Small-business owners and staff in customer interactions, back office, supply chain, and leadership. Partners: PayPal, Prospect Butcher Company (Brooklyn butcher shop), and MAKS TIPM Rebuilders (automotive power-module rebuilder). Context: limited resources, business accountability, customer experience. Modules: Intro & 4Ds → How AI works → Description-Discernment loop → Delegation-Diligence loop → Putting it together → Closure.
What’s specific:
- Values, goals, and constraints are written down up front so every AI interaction starts with the right context.
- Every efficiency gain should mean better customer service and more time for high-value work.
- Research loop (Mak, CEO of MAKS): use domain expertise to steer research. Verify regulations, pricing data, and deadlines against primary sources before acting.
- Data and transparency loop: decide which customer data to hand off, strip identifying details, choose tools whose privacy settings match the data’s sensitivity, and be transparent with customers about AI use. Act fast if something goes wrong.
- AI-augmented workflow: find repetitive tasks (e.g., FAQ responses) suitable for automation. Complaints and judgment calls stay human. Write Product, Process, and Performance Descriptions for the system (output, step-by-step logic, tone, boundaries, rules). Test with real examples, review before anything reaches customers, and always provide a clear path to a human.
- Human in the loop and AI use policy: a short, honest policy covering transparency, accountability, and AI dependency. Discuss what happens with saved time and protect the personal touch.
- Use blueprints for common business tasks and share your badge with your team and partners.
9. AI Fluency for Students
Who: Students using AI for learning and career preparation. Authors: Anthropic with Dakan & Feller. Driving question: How can I use AI to enhance my learning and career without compromising my growth and integrity? Modules: Intro & 4Ds → Applications (learning partner, career planning) → Being the human in the loop.
What’s specific:
- Augmentation over automation in learning: the aim is a partner that helps you do your work better, not a tool that does it for you.
- Personal learning context document to guide future AI collaborations.
- AI as a learning partner:
| D | For learning |
|---|---|
| Delegation | Know what you’re trying to learn; choose tools designed for education |
| Description | Steer the AI to act as a tutor or coach, not an answer-giver |
| Discernment | Be honest about whether you’re actually learning or just along for the ride |
| Diligence | Follow academic policies; be able to explain and defend everything you submit |
- Build an AI learning partner setup and a living learning journal to track growth. Real learning with AI takes more effort than shortcuts, and it pays off in exams and interviews.
- AI in career planning: career mapping (AI supplies information; you supply self-knowledge and values), skill building (AI as coach with specific goals and active practice), and job search (research and interview practice). Generic AI-written resumes and cover letters are easy to spot, so keep your authentic voice. AI Fluency is itself a career skill worth highlighting.
- Being the human in the loop: you decide what to ask, judge whether the output makes sense, and bring your values. Write a personal AI collaboration policy: when you will and won’t use AI, how you’ll use it for learning, transparency commitments, and how you’ll keep your skills sharp. Reflect regularly: Is this helping me grow? Am I being honest about AI’s role?
Quick-Review Flashcards
- Q: What are the two nested loops? → A: The outer Delegation ↔ Diligence loop (whether, when, and how to use AI, and owning the result) and the inner Description ↔ Discernment loop (describe, evaluate, refine).
- Q: Builders: what is the Description Chain? → A: User voice → requirement → technical spec → AI instruction → tests. Prompting is only one link.
- Q: Builders: what are the five Discernment lenses? → A: Correctness, quality, fit, experience, responsibility.
- Q: Creative: what are the three dimensions of the Creative Value Lens? → A: Creative Expression, Human Connection, Distinctiveness.
- Q: Creative: what are the five AI roles, and what line divides them? → A: No AI, Consultant, and Tool Builder vs. Co-Creator and Medium. The line is whether AI-generated material appears in the final work the audience encounters.
- Q: Creative: what happens to the Value Lens as production constraints fall? → A: It matters more, because purpose and specificity become what sets work apart.
- Q: pK–12: what data rule applies to AI tools? → A: Protections like FERPA still apply, so use only district-approved tools for student information.
- Q: Teaching AI Fluency: what are the four teaching approaches? → A: Linear, non-linear, focused, two-loops (loop-based).
- Q: Teaching AI Fluency: what are the three assessment approaches? → A: Outcome-based, process-based, reflection-based.
- Q: Nonprofits: how do you build trust in AI data analysis? → A: Test it on past data where you already know the answers, then document the approach that works.
- Q: Nonprofits and small businesses: what is the key Task Delegation question? → A: “Should AI do this?”, not just “can it?”. FAQs can be automated; complaints and judgment calls stay human.
- Q: Students: what is the main rule for anything you submit? → A: You must be able to explain and defend all of it, even if AI helped.
Practice Ideas
- Write your context document for your role (builder brief, teaching context, business values and constraints, or student learning context) and paste it at the start of your next three AI sessions.
- Builders: write 5 pass/fail acceptance tests and trace one feature through the full Description Chain before prompting.
- Creatives: score one recent piece on the Value Lens and Production Lens, then choose an AI role for each step and write a 3-question personal policy with a revision trigger.
- Educators: have AI help draft one differentiated material and check it for accuracy, reading level, and alignment. Then write a short AI value document.
- Teaching AI Fluency: design one assignment that has an outcome, a process (annotated chat log), and a reflection component.
- Trainers: do the learner workbook yourself and run a 20-minute pilot with two colleagues.
- Nonprofits and small businesses: list your 10 most frequent inbound questions, sort them into automate vs. keep human, and draft a one-page AI use policy.
- Students: set up a “tutor, not answer-giver” prompt for one course and keep a learning journal for two weeks.