What Is AI Fluency? The 4D Framework in Brief
Most conversations about AI at work start with tools: which model, which subscription, which prompt. Tools matter, but the bigger difference usually comes from the people using them. That skill has a name: AI fluency.
What AI fluency means
The AI Fluency Framework, developed by Professors Rick Dakan and Joseph Feller with Anthropic, defines AI fluency as interacting with AI systems in ways that are effective, efficient, ethical and safe. Notice the last two. Being fluent isn’t only about getting more done; it’s also about knowing when not to use AI, and owning what you do with it.
Three ways of working with AI
The framework describes three modes of interaction:
- Automation: AI executes specific tasks based on human instruction.
- Augmentation: humans and AI collaborate as thinking partners.
- Agency: humans configure AI to independently perform future tasks on their behalf.
None of these is “more advanced” than the others. The right mode depends on the task, and fluent teams switch between them deliberately.
The four competencies (the 4Ds)
- Delegation: “Setting goals and deciding whether, when and how to engage with AI.” Before opening a chat window, ask whether this is a task AI does well, and which parts you should keep for yourself.
- Description: “Effectively describing goals to prompt useful AI behaviors and outputs.” Give the context, constraints, examples, and format you would give a new colleague. Describe how you want the work done, not just what.
- Discernment: “Accurately assessing the usefulness of AI outputs and behaviors.” Fluent-sounding is not the same as correct. Check facts, reasoning, and whether the output actually serves your goal.
- Diligence: “Taking responsibility for what we do with AI and how we do it.” Decide what data you share, be transparent about the role AI played, and own the final result.
The competencies feed each other. Description and Discernment, in particular, work as a loop: describe, evaluate, refine, and repeat until the output is genuinely useful.
A simple way to start
You don’t need an AI strategy document to begin (I hate over-engineering). Pick one task your team does every week, such as summarizing a report or cleaning a dataset, and walk through the 4Ds together:
- Delegation: Which parts go to AI, and which stay with a person?
- Description: Write down the context and quality standards a new colleague would need.
- Discernment: Agree in advance how you will check the output.
- Diligence: Agree what data can be shared, and how you will disclose AI’s role.
One task, done well, teaches more than a dozen tools tried in passing.
Learn more
- Anthropic’s free course: AI Fluency: Framework & Foundations (I completed it; here is my certificate).
- Want help building AI fluency in your team or organization? See my Consulting page.
The AI Fluency Framework is © 2025 Rick Dakan, Joseph Feller, and Anthropic, released under the CC BY-NC-SA 4.0 license. The quoted definitions come from the framework’s summary.