Learn/Claude Academy: Become AI fluentguide

AI Fluency: Framework and foundations

The 4D framework of Delegation, Description, Discernment and Diligence.

claude-academyai-fluency4d-framework
Series Claude Academy: Become AI fluent Part 1 of 4
  1. AI Fluency: Framework and foundations
  2. AI capabilities and limitations
  3. Building effective human-agent teams
  4. AI Fluency for your world

A walkthrough of the Claude Academy course AI Fluency: Framework and foundations (Become AI fluent track, 14 lessons + quiz). 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 lesson summaries and key takeaways were used as the primary source; explanatory examples and tables are added study aids.)

Authors: Anthropic with Prof. Rick Dakan (Ringling College of Art and Design) and Prof. Joseph Feller (University College Cork), creators of the AI Fluency Framework (developed 2023–2024). Supported in part by the Higher Education Authority (Ireland). License: CC BY-NC-SA 4.0. Audience: Anyone at any level of AI experience who wants to collaborate with AI effectively and responsibly. This is the foundation course that every “AI Fluency for …” variant builds on. Format: 14 lessons (videos, exercises, reflections), a running personal project, a final quiz, and a completion badge.

Learning objectives (from the course page)

  • Explain what AI Fluency means and why it matters for collaborating with AI effectively, efficiently, ethically, and safely.
  • Apply the 4D Framework: Delegation, Description, Discernment, Diligence.
  • Distinguish Automation, Augmentation, and Agency as three ways of engaging with AI.
  • Describe how generative AI and LLMs work, including key capabilities and limitations.
  • Make Delegation decisions using Problem Awareness, Platform Awareness, and Task Delegation.
  • Communicate through Product, Process, and Performance Description plus foundational prompting techniques.
  • Evaluate outputs, processes, and behavior critically and refine results through the Description-Discernment loop.
  • Practice Creation, Transparency, and Deployment Diligence.

Module 1 — Introduction to AI Fluency

  • The course is about human-AI collaboration, not just learning about AI technology.
  • It aims for lasting skills instead of “cool prompts” or tricks that expire with the next model release.
  • AI Fluency = the ability to work with AI in ways that are:
Quality What it means in practice
Effective You get outcomes that meet your goals.
Efficient You save time and effort overall, including checking and fixing.
Ethical Your use is honest and fair and respects the people affected.
Safe You manage risks to data, people, and decisions.
  • Fluency combines practical skills, knowledge, insights, and values. It is a change in how you think about working with AI, not only a set of techniques.

Module 2 — The AI Fluency Framework

Why do we need AI Fluency? Three modes of interaction

Mode Definition (course wording) Your role Example
Automation AI executes specific tasks based on your instructions. Give the instruction, then check the result. “Reformat this list into a table.”
Augmentation You and AI collaborate as creative-thinking and task-execution partners. Thinking partner who iterates with the AI. Brainstorming and refining a presentation together.
Agency You configure AI to work independently on your behalf, shaping its knowledge and behavior patterns rather than specific actions. Designer/supervisor who sets up the AI’s knowledge, behavior, and boundaries. Setting up an AI tutor or assistant that interacts with others for you.
  • The 4Ds apply across all three modes. Developing them prepares you for AI capabilities that are still changing.

The 4D Framework at a glance

Competency Core question Sub-components
Delegation What work should be done by me, by AI, or together? Problem Awareness · Platform Awareness · Task Delegation
Description How do I communicate effectively with AI? Product · Process · Performance Description
Discernment Is what I got (and how I got it) any good? Product · Process · Performance Discernment
Diligence Am I using AI responsibly and owning the result? Creation · Transparency · Deployment Diligence
  • The 4Ds are interconnected rather than strictly sequential. Description and Discernment form a tight loop. Delegation and Diligence frame the strategic and ethical “should we and how?” (other courses call these the two nested loops).

Module 3 — Deep Dive 1: What is Generative AI?

Generative AI fundamentals

  • Generative AI creates new content (text, images, code) instead of only classifying or analyzing existing data.
  • Large language models (LLMs) such as Claude were made possible by three developments:
    1. Algorithmic/architectural breakthroughs, especially the transformer architecture
    2. Large amounts of digital training data
    3. Large increases in computational power
  • Training happens in two stages:
    • Pre-training: the model learns patterns from billions of text examples.
    • Fine-tuning: the model learns to follow instructions and give helpful responses.
  • Other concepts covered: the context window (how much the model can attend to at once) and emergent capabilities.

Capabilities & limitations

Capabilities Limitations
Versatile across many language tasks Knowledge cutoff (no knowledge after its training data unless tools provide it)
Keeps conversational context and flow Hallucinations: confident but incorrect outputs
Switches between diverse tasks without retraining Context window constraints
Can connect to external tools Difficulty with complex reasoning
  • The field changes quickly. The best results combine AI’s speed and scale with human critical thinking, judgment, creativity, and ethical oversight.

Module 4 — Delegation

  • Delegation means deciding what to do yourself, what to do with AI, and what to let AI handle, then distributing the work across those three options.
Sub-competency Meaning Guiding questions
Problem Awareness Understanding your goals and the work needed before you involve AI What is the vision? What does a good result look like?
Platform Awareness Knowing what different AI systems can and can’t do Which tool fits the task? What are its limits and data policies?
Task Delegation Dividing work strategically between you and AI to use each one’s strengths Which parts need human expertise, creativity, or judgment?
  • Good Delegation needs domain expertise and an understanding of AI capabilities.
  • The goal is the most effective human-AI partnership for the task, not automating everything.
  • Project planning exercise: choose a medium-sized, multi-step project of about an hour (communication, research, creative, or learning). Work with Claude to define the vision, break it into tasks, and write a delegation plan. You use this project in later lessons.

Module 5 — Description

  • Description means communicating with AI in a way that sets up a productive collaborative environment. It is more than writing a single prompt.
  • AI is an interactive partner, not a database or vending machine. It can’t read your mind.
Type What you specify Example phrasing
Product Description What you want: output, format, audience, style “A one-page brief for non-technical managers, bullet format.”
Process Description How the AI should approach the task “First list assumptions, then compare three options, then recommend.”
Performance Description How the AI should behave during the collaboration “Be concise; challenge my reasoning rather than agreeing.”

Deep Dive 2: Effective prompting techniques

  • Prompt engineering means designing effective instructions. It combines ordinary communication principles with AI-specific techniques.
  • Six foundational techniques:
    1. Give context: what you want, why you want it, and relevant background.
    2. Show examples of the style or format you want.
    3. Specify constraints: format, length, and other requirements.
    4. Break complex tasks into steps.
    5. Ask the AI to think first before it answers.
    6. Define the AI’s role or tone.
  • “Secret weapon”: ask the AI to help improve your prompt.
  • Prompting is iterative and can be done collaboratively with the AI. When a response falls short, refine the prompt instead of starting over.

Module 6 — Discernment

  • Discernment means evaluating what AI produces, how it got there, and how it behaves. It is the counterpart to Description.
Type Evaluates Look for
Product Discernment Quality of the output Accuracy, appropriateness, coherence, relevance
Process Discernment How the AI reached the output Logical errors, gaps in attention, inappropriate reasoning
Performance Discernment How the AI behaved in the interaction Whether its communication style suits your needs (e.g., too agreeable or too verbose)
  • Even the most advanced systems need human judgment and oversight.

The Description-Discernment loop

  • Describe → AI responds → Discern → Describe again more precisely. Each round of evaluation shapes the next instruction.
  • Applied to your course project, the goal is a result that neither you nor the AI would have produced alone.

Module 7 — Diligence

  • Diligence means taking responsibility for your AI collaborations. It covers the ethical and safety side, while the other Ds focus mainly on effectiveness and efficiency.
Type Meaning Practical checks
Creation Diligence Choosing AI systems thoughtfully and using them with care Is this tool appropriate for the data and task? What are its risks?
Transparency Diligence Being honest about AI’s role with everyone who needs to know Disclose AI use to teachers, clients, or colleagues as expected.
Deployment Diligence Taking ownership of AI-assisted outputs you use or share Verify facts, check for bias, and be ready to vouch for the result.
  • Expectations for disclosure and verification differ across personal, academic, and professional contexts. You are responsible for knowing and meeting them.
  • The course demonstrates this with an AI diligence statement that explains how Claude helped build the course while the human authors made all final decisions.

Module 8 — Conclusion, quiz & additional activities

  • Fluency comes from practice over time, not from mastering it all at once.
  • AI is powerful but not magic. It is only as useful and safe as the way we engage with it.
  • The framework is designed to remain relevant as models change.
  • Additional activities: build a personal AI fluency plan (rate each D as novice, developing, or confident and choose 1–2 to focus on), build a personal prompt and pattern library (5–10 reusable templates), and “Game night” word puzzles to practice precise Description and Discernment.

Quick-Review Flashcards

  1. Q: What are the four qualities of AI Fluency? → A: Effective, efficient, ethical, and safe collaboration with AI.
  2. Q: Name the 4Ds. → A: Delegation, Description, Discernment, Diligence.
  3. Q: Automation vs. Augmentation vs. Agency? → A: In Automation, AI does specific tasks you instruct. In Augmentation, you and AI think and create together. In Agency, you configure AI’s knowledge and behavior so it can act independently on your behalf.
  4. Q: Three components of Delegation? → A: Problem Awareness, Platform Awareness, Task Delegation.
  5. Q: Product vs. Process vs. Performance Description? → A: What you want · how the AI should approach it · how the AI should behave during the collaboration.
  6. Q: What does Process Discernment look for? → A: Flaws in how the AI reached its answer, such as logical errors, gaps in attention, or inappropriate reasoning.
  7. Q: Three types of Diligence? → A: Creation (choice of tool and how you use it), Transparency (disclosing AI’s role), Deployment (verifying and owning outputs).
  8. Q: Three developments that enabled LLMs? → A: The transformer architecture, large training datasets, and large increases in compute.
  9. Q: Two training stages? → A: Pre-training (learning patterns) and fine-tuning (learning to follow instructions helpfully).
  10. Q: Name four key limitations of current generative AI. → A: Knowledge cutoff, hallucinations, context window limits, and difficulty with complex reasoning.
  11. Q: List the six prompting techniques. → A: Give context, show examples, specify constraints, break tasks into steps, ask the AI to think first, define role or tone.
  12. Q: What is the Description-Discernment loop? → A: An iterative cycle where you evaluate the output and use that evaluation to write a better next instruction.

Practice Ideas

  • Delegation plan: choose a real task and list its sub-tasks. Label each one Human / Together / AI and choose a mode (Automation, Augmentation, or Agency).
  • Three-P prompt rewrite: take a vague prompt you’ve used and rewrite it with explicit Product, Process, and Performance Description. Compare the results.
  • Discernment audit: for one AI answer, list one Product, one Process, and one Performance observation. Then feed them back to the AI and see whether the output improves.
  • Hallucination hunt: ask for sources on a niche topic and check each one.
  • Diligence statement: write a 3–4 sentence disclosure of AI’s role in a recent piece of work.
  • Prompt library: save 5 templates for recurring tasks, with placeholders and notes on what makes each one work.