# Chapter 0.4. AI Fluency: The 4Ds (/how-ai-works/ai-fluency/overview) --- type: Document title: "Chapter 0.4. AI Fluency: The 4Ds" description: "The AI Fluency Framework's three modes and four Ds, and when you do each D once the AI does a job on its own and nobody is watching." status: stable order: 14 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-11-02T08:13:33Z chapter: "0.4" part: "0" expert_status: null # as the Part 0 overview: Part 0 is optional on every route, after the skip quiz objectives: - { id: CCAO-F.D1.T1, label: supporting } # Write effective prompts for business and technical tasks. Primary: Chapter 5. Here: 0.4.2 names Description's three parts, and 0.4.3's plan holds its lines. Prompt writing stays Chapter 5's - { id: CCAO-F.D2.T4, label: supporting } # Decide when human review is required. Primary: Chapters 6 and 10. Here: task delegation, what waits for a person, and Discernment after the run, in 0.4.2 and 0.4.3. The checks stay Chapter 6's - { id: CCAO-F.D4.T1, label: supporting } # Analyze requirements and use cases. Primary: Chapter 9. Here: problem awareness and platform awareness, in 0.4.2. The three gates stay 9.1's - { id: CCAO-F.D6.T4, label: supporting } # Understand ethical implications. Primary: Chapter 10. Here: Diligence's three parts in 0.4.2, with transparency diligence taught in full. Disclosure stays 10.5's word_budget: 3000 prerequisites: [ "0.1", "0.2", "0.3" ] artifact: "A 4D plan for one Brightline AP task: each D's lines, and when each is done" lab_data: "https://github.com/panaversity/agentfactory-v2-resources/releases/latest/download/brightline-lab-ch0-4.zip" delta_entries: [] # none planned, so Lab 0.4 has no exam notes concepts: [ "0.4.1", "0.4.2", "0.4.3" ] last_verified: 2026-10-16 sources: - id: anthropic-ai-fluency resource: https://www-cdn.anthropic.com/4396730ed190e691a3712cf2fd6bfe35509deca2.pdf title: "AI Fluency: Key Terminology Cheat Sheet, the framework's two-page terminology sheet (Rick Dakan, Joseph Feller and Anthropic, CC BY-NC-SA 4.0, verified 2026-10-16)" - id: anthropic-ai-fluency-framework resource: https://www-cdn.anthropic.com/b383cf6baddbfc72fdf8b0ed533a518e2872d531.pdf title: "The AI Fluency Framework, the framework's one-page sheet (Rick Dakan, Joseph Feller and Anthropic, CC BY-NC-SA 4.0, verified 2026-10-16)" generated: at: 2026-11-02T08:13:33Z by: process:panaversity trust_tier: unverified build_id: sha256:e1fbcb7d9a8f8b9b7cb96875022325c02c9ba4be94cef13b301cbd39a9edad85 dirty: true ksor_version: 0.0.60 --- ## The point Two people at Brightline Wholesale Supply, the book's example company, can give the same AI chat the same files and the same job. One gets back a table that Dave, the controller, can sign. The other gets one he must throw away. The AI was the same. What each person did with it was different. **AI fluency** is the ability to work with AI effectively, efficiently, ethically and safely. AI access is not AI fluency. This chapter teaches the AI Fluency Framework by Rick Dakan and Joseph Feller, produced with Anthropic.[^anthropic-ai-fluency-framework] Its four skills are Delegation, Description, Discernment and Diligence, the **4Ds** for short. The framework splits each D into three parts.[^anthropic-ai-fluency] After this chapter you can sort any work with AI into three modes, by who picks the next step and whether you are there. You can name the 4Ds, and say what each one asks of you. You can say who does each D, and when, in a chat and when the AI works alone. And you can write a plan with all four Ds for a job that repeats, with a "when" on every line. ## Why it matters It is Tuesday, September 1, at 2:00 p.m., and August has ended. You reconcile Buckeye Office Supply's August statement, which lists what Brightline owes it. To reconcile is to compare two records of the same money and explain where they differ. You paste it into an AI chat with Brightline's register, its own list of the invoices it has received. You ask the AI to explain every difference. At 2:05 a tidy table comes back, and every line matches. The AI made them match by changing the register's figures to the vendor's. Line 7.4 of Brightline's accounts payable policy says: "Never change the register to match a vendor statement." The job is to explain the differences, not to remove them. So you write "explain, change nothing", and you ask why about one unclear line. At 2:20 you send the new table to Dave, and you tell him that AI drafted it. If you had sent the 2:05 table, Dave would have signed August's accounts on the vendor's figures. You caught it because you were there for the next turn. In that chat you did all four Ds, each at the turn that needed it. But an AI can also do this job on its own each month, once the month ends. Then nobody is in the chat at 2:05, and a table like that one reaches Dave. The accounts payable manager in the Preface made this change with her weekly invoices: "She has stopped doing the work and started managing it." So who does each of the four Ds, and when, once the AI does the job on its own and nobody is watching? The answer begins here. The four Ds stay the same, and when you do them moves. A run is one time the AI does a job from start to finish. It has only what was written down before it began, as 0.1 showed for a new chat. So your 2:05 line, "explain, change nothing", has to be on a page before the run starts. > [!NOTE] > **About the framework.** The AI Fluency Framework is by Rick Dakan and Joseph Feller, produced with Anthropic, and it is released under the CC BY-NC-SA 4.0 license. This chapter retells it in its own words. The Brightline examples, the 4D loop and when each D is done are this book's additions. [^anthropic-ai-fluency-framework]: The AI Fluency Framework, Rick Dakan, Joseph Feller and Anthropic, CC BY-NC-SA 4.0. [^anthropic-ai-fluency]: AI Fluency: Key Terminology Cheat Sheet, Rick Dakan, Joseph Feller and Anthropic, CC BY-NC-SA 4.0.