Everyone in health IT these days seems to be “in AI.” It’s the buzzword on every badge, every product roadmap, and nearly every panel discussion. But let’s be honest: most people talking about AI will admit they don’t really understand what it is or how it works. This piece is for the curious, the skeptical, and the serious professionals who want to go beyond the surface.
Let’s start with a familiar experience.
Everyone has that one friend—you know the one. You start talking, and they cut you off halfway through your sentence. Maybe it’s because they’re impatient. But more often, it’s because they think they know where you’re going. They’ve already filled in the rest of your thought before you finish. Sometimes they’re right. Sometimes, not so much.
But here’s the thing—your friend isn’t the only one who does this. You do it too. We all do.
Let’s see how good you are at filling in the rest. Try these:
- Nike: Just Do ____.
- Snap, Crackle, ____.
- The Few. The Proud. The ____.
Your brain fills in the blanks almost instantly. Not because you stopped to think, but because you’ve seen these patterns so many times before. That’s pattern recognition. That’s probabilistic reasoning. That’s guessing—smart, fast, subconscious guessing.
And that’s exactly how today’s most advanced AI systems work.
So, what is AI?
According to the National Institute of Standards and Technology (NIST)—the U.S. federal agency responsible for setting technical and cybersecurity standards—AI is defined as:
“A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments.” (NIST SP 800-218)
It’s a clean, concise definition. But what does that mean in practice?
Let’s break it down.
At the core of generative AI—like ChatGPT—is a process called probabilistic pattern matching. These models don’t “think” like humans. They don’t “understand” meaning the way we do. Instead, they break down your input into tokens—small chunks of language—and then guess what should come next, based on patterns they’ve seen before.
For example, consider the sentence: “AI will transform healthcare.”
This may get tokenized like:
- “AI”
- “ will”
- “ transform”
- “ health”
- “ care”
- “.”
Each token is a unit of language—sometimes a full word, sometimes part of one. And after reading those, the model doesn’t reflect on meaning or ethics or intent. It just starts writing. One token at a time. It guessesor predicts the next token based on probabilities, then the next, and the next—until it has built a full sentence, paragraph, or page.
Still not convinced it’s just pattern recognition? Try this.
Here’s a fun prompt engineering exercise. (That’s just a fancy term for “what you type into the box.”)
In your favorite generative AI platform—ChatGPT, Claude, Gemini, whatever—try typing: “What is AI?”
Now try: “Wtha si AI?”
Even with the typo, the model guesses correctly. Why? Because it’s seen enough examples—of both clean and messy inputs—to recognize what’s probably intended. It doesn’t understand like a person. It’s just making really smart, context-aware guesses. Over and over.
But AI isn’t magic. It’s math.
And I get it. You probably hate math. Most people do. But I love it.
Ask me about Organic Chemistry, though? Total disaster. (Fun moment: my daughter’s new boyfriend recently asked what my hardest subject was in college. Without hesitation: Org Chem. Hands down.)
That love of math, though—that’s what makes this space so exciting to me. Because once you get past the buzzwords and the black-box fear, AI is just probability applied at massive scale. It’s not mysterious. It’s not even particularly “intelligent.” It’s a powerful engine for recognizing and recreating patterns—one token at a time.
And in health care, that distinction matters. A human doctor makes decisions rooted in uncertainty, ethics, and context. An AI model makes predictions based on patterns. So when we talk about using AI in clinical care, documentation, diagnostics, or administrative workflows, we must be clear-eyed: AI is a tool. A powerful one, yes—but a tool that reflects the data it was trained on, and the intent with which it’s deployed.
So what is AI really?
It’s a guessing game. A really, really powerful guessing game built on lots and lots of prior examples.
How will AI change the health IT world?
Well… Your guess is as good as mine.
DISCLAIMER – The views expressed in this post are solely my own and do not represent the views of my employer, clients, or any affiliated organizations. This is not intended as legal, financial, or professional advice. Any opinions shared are personal reflections meant to foster discussion and engagement.
