Four Ways to Define AI — Find the One That Clicks for You

Before my kids left for college, I gave them a study tip I learned the hard way: don’t just read one textbook definition of a concept. Go to the library. Find three or four textbooks. Read how different professors define the same idea. One of them will click — because of a metaphor, a tone, a way of framing it. That’s the moment: “Now I get it.” Abstract concepts need that moment. I was a physics major. That’s how I survived.

When I wrote my own textbook, I wanted to create those moments for AI. In Chapter One, I tried to define AI four different ways — technologically, conceptually, ethically, and by role. These are my own attempts, not established standards — but I hope one of them clicks for someone the way a good textbook definition once clicked for me.

Most AI definitions are written for engineers. They describe algorithms, data structures, and training pipelines. Useful if you’re building AI. Less useful if you’re trying to understand what AI does in the world.

Here’s one way I’ve tried — a role-based framing I find useful, offered as a starting point, not a verdict:

AI is a field of computer science that designs systems to amplify human problem-solving, created by builders and guided by bridges, to perform reasoning, prediction, or action at scale.

Two words carry the weight: builders and bridges.

Builders are the technologists — engineers, data scientists, and developers who design AI systems. From their perspective, AI is a collection of algorithms and data structures that recognize patterns, learn from experience, and make decisions.

Bridges are domain experts — people who understand a field deeply enough to translate human needs into something AI can act on. From their perspective, AI is a thinking partner. It helps analyze complexity, test options, and accelerate insight. But it still needs human values, context, and judgment to be useful.

This framing isn’t about technical skill. It’s about how you relate to AI — whether you’re constructing it or connecting it to problems that matter.

The Observer’s Insight

Most people skip definitions and jump straight to tools. I understand why — tools are immediate, definitions feel abstract. But how you define AI shapes what you expect from it, what questions you ask of it, and what you demand of yourself when working with it.

A definition is not just a sentence. It’s the foundation of your entire mental model. When that foundation is shaky — when you’re working from a vague or borrowed understanding — everything built on top of it wobbles too.

I’ve seen this up close in my own teaching. When a student struggles with an assignment, the problem is rarely the assignment itself. More often, it’s a gap several steps back — a key concept that was never fully grasped, just carried forward. You can’t build on ground that isn’t solid. And the further you go, the harder it becomes to go back and fix what was missed.

There is another reason definitions matter. A definition is portable in a way that specific tools and theories rarely are. Because it operates at the level of fundamentals, it travels across situations — you can use it to recognize AI when you encounter it, name what it’s doing, and evaluate whether it’s being applied well. Specific tools, by contrast, are narrower. They solve particular problems in particular contexts, and they don’t transfer easily beyond those boundaries.

Definitional fluency is what gives you judgment across all of them.

The right definition doesn’t just explain a concept. It organizes everything else you learn around it. That’s what I was trying to do in Chapter One — not to settle what AI is once and for all, but to offer a few angles, and let readers find the one that gives their knowledge somewhere solid to stand.

If the role-based definition doesn’t click, there are three others waiting.

The eBook version of my textbook will be available beginning April 16 through Springer. If you’re a student or faculty member at a university that licenses the Springer Nature eBook Collection, the book is free — just search “Artificial Intelligence in Public Health” through your university library or SpringerLink. Chapter One alone covers multiple AI definitions.

Which framing of AI has clicked for you — or are you still looking for that moment? Leave a comment or reply directly — I’d love to hear.

— Min Wu, PhD Associate Professor,
Zilber College of Public Health
University of Wisconsin-Milwaukee