November 9, 2026 · 2 min read
What Is AI Fluency? A Plain-English Guide for Teams
AI fluency isn't knowing every tool. It's four habits that make any tool useful. Here's what they are.
AI fluency is the ability to work with AI effectively, efficiently, ethically, and safely, whatever the tool. A widely used way to teach it is the AI Fluency Framework by Rick Dakan and Joseph Feller, developed with Anthropic, which breaks it into four skills: Delegation, Description, Discernment, and Diligence.
Why tool training isn't enough
Most AI training teaches buttons: where the features are and what to click. Features change every few months. The skills that last are knowing what to hand to AI, how to ask for it, how to judge what comes back, and how to use it responsibly.
The four skills
- Delegation: deciding what work to do with AI and what to keep doing yourself, based on the task, the stakes, and what the tool is good at.
- Description: communicating clearly with AI, including the goal, the context, the format you want, and how you want it to work.
- Discernment: critically evaluating what AI produces, how it got there, and whether it's right, in proportion to what's at stake.
- Diligence: taking responsibility for how you use AI, being open about it where it matters, and owning the result.
Three ways of working with AI
The framework also describes three modes, and fluent teams move between them:
- Automation: AI carries out a specific task you've defined.
- Augmentation: you and AI work together as thinking partners.
- Agency: you set AI up to work independently on your behalf.
Which skill to work on first
Most teams get better at Description just by using AI more. Discernment doesn't improve on its own, and it's the skill that prevents mistakes from reaching customers. That's why it's worth practicing on purpose: check AI output against what you know, ask how it reached an answer, and look harder when the stakes are higher.
How to build AI fluency on a team
- Practice on real work, not demos.
- Agree on simple rules for data and review, so people aren't guessing.
- Share what works: prompts, examples, and mistakes worth avoiding.
- Revisit it as tools change, since the habits stay the same even when the buttons move.
Questions people ask
01Who created the AI Fluency Framework?
Rick Dakan and Joseph Feller, developed with Anthropic. Course materials are published under a Creative Commons license (CC BY-NC-SA 4.0).
02Does AI fluency mean learning to code?
No. It's for everyone who uses AI at work. The four skills apply to writing an email with AI as much as to building software with it.
03How long does it take a team to become AI fluent?
The basics can be taught in a hands-on session, but fluency comes from practice. Teams improve fastest when they use AI on real work and review the results together.
Next article5 Common Mistakes Businesses Make When Adopting AI (and How to Avoid Them)
