AI Fluency: The 4D Framework · lesson 1 of 2

AI Fluency: The 4D Framework

What the course is

AI Fluency — Anthropic's flagship course on working with AI effectively, efficiently, ethically, and safely, co-authored with Prof. Joseph Feller (University College Cork) and Prof. Rick Dakan (Ringling College of Art and Design). It's deliberately tool-agnostic: the framework applies to Claude but also to any generative AI system.

The core claim: AI fluency is not prompting tricks. It's four durable competencies — the 4Ds — that stay relevant as models change.

The 4D Framework

Delegation — deciding what work to give AI in the first place. Covers problem awareness (what am I actually trying to achieve?), platform awareness (what is this AI good/bad at?), and task delegation (splitting work between human and AI). Good delegation is project planning: decompose the goal, then assign each piece to whoever — human or model — does it best.
Description — communicating what you want clearly enough that the AI can deliver it. This is where prompting techniques live: providing context, specifying output format, defining audience and constraints, iterating on phrasing. Description is a conversation skill, not a one-shot incantation.
Discernment — evaluating AI output critically rather than accepting it. Assessing the product (is it correct, complete, appropriate?), the process (did the AI reason sensibly?), and the performance (is this collaboration working?). Discernment is what catches hallucinations, subtle errors, and plausible-but-wrong answers.
Diligence — taking responsibility for how you use AI: being transparent about AI involvement, staying accountable for what you ship, and using AI in ways consistent with your obligations and values. Diligence covers creation diligence (choosing appropriate tools/uses), transparency diligence (disclosure), and deployment diligence (owning the output).

The two loops

The 4Ds pair into two feedback loops worth memorizing:

Loop	Pairing	The cycle
Description–Discernment loop	Describe → evaluate → re-describe	The tight interaction loop: every prompt is a hypothesis; the output tells you how to refine it. Most day-to-day skill lives here.
Delegation–Diligence loop	Delegate → stay accountable → re-scope	The outer judgment loop: what you choose to hand off determines what you must verify and own. High-stakes tasks demand narrower delegation and heavier diligence.

A useful summary: Description and Discernment are how you work with AI; Delegation and Diligence are how you decide whether and how AI should be involved at all.

Audience variants

Anthropic ships the framework in tailored editions, all built on the same 4D core:

AI Fluency for Educators — course design, academic integrity, modeling fluency for students.
AI Fluency for Students — study workflows, learning with AI without outsourcing understanding.
Teaching AI Fluency — a train-the-trainer edition for people delivering the framework to others.
AI Fluency for Nonprofits — co-created with GivingTuesday; resource-constrained, mission-driven contexts.
AI Fluency for Small Businesses — owner-operator workflows, customer-facing use.
AI Fluency for Builders — the developer edition, bridging into API work (claude-platform-101).
pK-12 Educators learning path — a structured path for primary/secondary educators.

Why it matters for cert prep