Sharpwork by Chris Fildes

AI Fluency Framework: Engineering With AI Starts With Being More Human, Not Less

Chris Fildes · · 4 min read

Being human matters even more in the era of AI

AI is changing the way we design, build, and run software. Yet as AI gets better, the human side of engineering is actually becoming more important.

AI can generate code, analyse patterns, and automate workflows, but it cannot replace the human abilities that make engineering safe, creative, and meaningful. Skills like judgment, context, creativity, and oversight are not extras; they are the foundation of working well with AI.

This is where the AI Fluency Framework comes in. Its key skills give us a clear way to use our human strengths in a world shaped by AI.

What the AI Fluency Framework Is - and Where It Comes From

The AI Fluency Framework is a model for how people and AI work together, created by Prof Rick Dakan (Ringling College of Art and Design) and Prof Joseph Feller (University College Cork). It defines AI Fluency as the ability to work well, efficiently, ethically, and safely with new ways of using AI, such as automation, agency, and augmentation.

The framework came from their research on how people really work with AI in real situations. As AI tools improved, Dakan and Feller saw that the main challenges were not technical, but human. People needed lasting skills to help them work with AI, no matter which tools or technologies they used.

Their research found four key skills, called the 4Ds, that lead to successful, safe, and meaningful work with AI:

  • Delegation: deciding what work to give AI
  • Description: communicating clearly and contextually
  • Discernment: evaluating outputs and reasoning
  • Diligence: ensuring responsible, safe, accountable use

Organizations have adopted and built on this model because it gives a human-centered, lasting foundation for working with AI at any skill level.

Let's take a closer look at each key skill, why it matters, and how it appears in real engineering work.

1. Delegation - Human Judgment in Action

What it is: Delegation is the skill of deciding what work to do yourself, what to do with AI, and what to let AI handle independently.

Why it matters: AI is powerful, but it cannot do everything. Delegation helps us use AI where it adds value and avoid it where risk, nuance, or special knowledge are needed. Poor delegation can mean relying too much on AI, not using it enough, or using it in unsafe ways.

Good delegation: The engineer writes the core logic and uses AI to generate test scaffolding.

Bad delegation: The engineer lets AI generate the entire feature, including business-critical logic, without review.

Delegation lets people use their judgment to decide how AI should be used.

2. Description - Human Context Made Explicit

What it is: Description is how we communicate with AI by setting goals, limits, examples, tone, and process.

Why it matters: AI does not pick up context like people do. Giving a clear description helps avoid confusion, cuts down on rework, and makes sure the AI understands what you really want, not just what you wrote.

Real-world example: You ask AI to “write documentation for the API.”

Weak description: “Write API docs.”

Strong description: “Write API docs for junior engineers joining the team. Focus on authentication, rate limits, and error handling. Use concise examples and avoid jargon.”

The difference is clear because people provide the context that AI cannot create.

3. Discernment - Human Critical Thinking

What it is: Discernment means being able to judge AI outputs, reasoning, and behaviour.

Why it matters: AI can give smooth, confident answers, even when they are wrong. Discernment helps keep mistakes, wrong assumptions, and hidden logic errors out of real systems.

Real-world example: AI generates a migration script for a database.

A discerning engineer checks:

  • Does it handle null values?
  • Does it lock tables?
  • Does it break referential integrity?
  • Does it scale for production data volumes?

Discernment is a form of oversight that makes sure AI's contributions are correct and trustworthy.

4. Diligence - Human Responsibility

What it is: Diligence is how people stay responsible for work done with AI by checking, being open, and making sure it is used safely.

Why it matters: AI can help, but people are still responsible for the results. Diligence makes sure we use AI ethically, openly, and reliably.

Real-world example: A product manager uses AI to summarise customer feedback.

Diligence means:

  • verifying the summary against real data
  • disclosing AI involvement when sharing insights
  • ensuring no sensitive information was mishandled
  • validating that the AI didn't introduce bias

Diligence is about integrity. It is the human part of working responsibly with AI.

Conclusion: Human Skills + AI Fluency = Sustainable Engineering

The 4D framework does not replace the basics of engineering. Instead, it makes them stronger. It helps us build systems that are:

  • more maintainable
  • more understandable
  • more resilient
  • more aligned with human values

AI speeds up the work, but people decide where it goes. The best results come from mixing human strengths with AI abilities, not by seeing AI as a replacement for human thinking.