AI Fluency: the 4 skills that matter
Tricks change every month, skills last. Meet the 4Ds for working well with AI.
Online you’ll find thousands of “tricks” and “magic prompts”. The problem? Assistants change fast, and what worked six months ago may be useless today.
That’s why this course is built on a sturdier idea: AI FluencyThe ability to work with AI in ways that are effective, efficient, ethical and safe. It’s not about knowing the tech, but about working well with it., the ability to work with AI in a way that is:
- effective → you get what you really need;
- efficient → without losing more time than you save;
- ethical → honest, towards others and yourself;
- safe → protecting your data and other people’s.
This lesson is inspired by the AI Fluency Framework by Rick Dakan and Joseph Feller, also used in Anthropic Academy courses. Here it’s explained simply, with everyday examples.
Think of AI as a collaborator
Imagine a new colleague: super fast, available at any hour, has read everything… but doesn’t know you, knows nothing about your situation and now and then says wrong things with great confidence.
To work well with them you need four skills. These are the 4Ds:
The 4Ds aren’t steps in a strict order: they intertwine. But if you ask yourself their four questions, you’ll avoid almost all the common mistakes.
Three ways of working with AI
Not all collaborations are the same. The framework describes three modes:
🤖 Automation — the AI carries out a specific task you gave it. “Translate this email into Italian.” You give the order, it executes.
🤝 Augmentation — you and the AI think together, like two colleagues. “Help me figure out if this mortgage is right for me: ask me questions.” The result comes from the back-and-forth.
🧭 Agency — you set up the AI to work on its own on your behalf, maybe talking to other people or using other programs. “Create an assistant that answers my B&B guests’ frequent questions.” Here diligence matters even more.
Why it’s worth learning
People who use AI without a method usually end up at one of two extremes:
- Blind trust: they copy-paste everything, and sooner or later get burned by a mistake.
- Total rejection: after one disappointing answer they decide it’s “useless”.
The 4Ds keep you in the middle: you use AI where it’s strong, check it where it’s weak, and you stay in charge.