I get asked some version of “what is AI?” probably three or four times a week. And honestly, I don’t blame anyone for asking, because the way AI gets talked about in the media and by vendors makes it sound like it’s one big thing. It’s not. Not even close.
AI is a whole category of technologies, and lumping them all together is like calling a truck, a sports car, and a minivan the same vehicle because they all have four wheels. They share some basics, sure, but they do very different jobs.
I put this series together because I think business leaders deserve a straight explanation, not a buzzword-heavy pitch. Part 1 is what you’re reading now. We’re going to dig into what AI actually is, where it came from, and the different types out there. Part 2 is where we talk about how to actually put it to work for your business. But you’ve got to know what you’re dealing with first, so let’s start there.
A (Very) Brief History of AI
The concept of artificial intelligence goes back further than most people expect. In the 1950s, researchers started asking a deceptively simple question: can machines think? Alan Turing famously proposed a test for it, and computer scientists spent the next several decades trying to answer it.
AI has gone through a few rounds of hype and disappointment since then. In the 1980s there was a surge of interest and funding, followed by what’s called an “AI winter” when the technology couldn’t live up to expectations. It happened again in the early 2000s. Each time, researchers kept working, the underlying technology kept improving, and the next wave came back stronger.
I actually watched this unfold from inside higher education. During my time working in academia and finishing my Masters in Learning Design and Technology, Machine Learning was already being applied to education. I co-authored a paper in 2018 on what we were calling “adaptive learning,” which was essentially AI-driven systems that could adjust how content was delivered based on how individual students were learning. The technology was already there. Most people just weren’t paying attention to it yet.
The most recent wave hit in late 2022 when OpenAI released ChatGPT. Suddenly AI wasn’t something only tech companies and academics talked about. It was everywhere overnight. But here’s what I think is important to understand: ChatGPT didn’t invent AI. It made one specific type of AI accessible to regular people, and that made the whole category feel brand new when it really wasn’t. A lot of the technology powering today’s AI tools had been in development for years before that. ChatGPT just opened the door for everyone to walk through it.
Understanding that distinction is going to matter a lot as we work through this.
What AI Actually Is (And Isn’t)
At its core, AI is software that can learn patterns, make decisions, or produce output without being hand-programmed for every possible scenario. A traditional computer program does exactly what you tell it to do, step by step. AI flips that around. You give it data and a goal, and it figures out how to get there.
That sounds impressive, and in many ways it is. But it’s worth being clear about what AI isn’t. It’s not sentient. It doesn’t understand things the way you do. It’s not a search engine, even though some AI tools feel a lot like one. And it’s definitely not one single technology. It’s a broad family of approaches, and the differences between them matter, especially when you’re making business decisions about which tools to invest in.
The Three Main Types of AI
When people talk about AI in a business context, they’re usually referring to one of three broad categories. Knowing which is which will save you a lot of confusion.
Machine Learning is the oldest of the three and probably the most quietly useful. Machine Learning is AI that learns from large amounts of data to spot patterns and make predictions. It’s been running behind the scenes for years. When your email filters out spam, when Netflix recommends something you actually want to watch, when your bank flags a suspicious transaction, that’s Machine Learning at work. It doesn’t create anything new. It analyzes what already exists and gets better at it over time. I work with this type of AI regularly, and it’s often the right fit for businesses that need to make smarter predictions without a lot of fanfare.
Generative AI is the flashy one, and it’s probably what most people picture when they hear “AI” these days. This is the type that creates new content: text, images, code, music, video. ChatGPT, DALL-E, and tools like them fall into this category. A lot of people assume Generative AI is actually “thinking” or “understanding” when it produces something. It’s not, really. It’s doing something closer to very sophisticated pattern completion. It’s learned from enormous amounts of existing content and can produce new combinations that look remarkably human. That’s genuinely impressive, but it also means it can confidently produce things that are wrong. I see that happen a lot when people use these tools without knowing the limitations.
Agentic AI is the newest of the three and the one I’m most excited about from a business perspective. Agentic AI can take actions, make decisions, and work through multi-step tasks without a human guiding it at every turn. Think of it less like a tool you type questions into and more like a system that can actually go do something for you. This is where I see the biggest opportunities for small and midsize businesses right now, and we’ll get into why in Part 2.
AI Models: The Engines Behind the Tools
Every AI tool runs on what’s called a “model.” A model is basically the brain of the operation. It’s the thing that was trained on data and learned to do a specific job. The tool is the interface you interact with. The model is what’s actually doing the work.
There are a few main types of models worth knowing about.
Large Language Models (LLMs) are what power most of the Generative AI tools you’ve probably already used. ChatGPT, Claude, Google’s Gemini, they all run on LLMs. These models were trained on massive amounts of text and learned to predict and generate language. They’re good at writing, summarizing, answering questions, and working with text in general. They also have real limitations, which we’ll touch on.
Predictive and Statistical Models are the engines behind Machine Learning. These models are trained on specific datasets to forecast outcomes or classify information. A model that predicts whether a customer is likely to churn, or which products will sell well next quarter, is this type. They’re less flashy than LLMs but often more immediately valuable for specific business problems.
Autonomous Agent Models are what power Agentic AI. These are newer and more complex. They’re not just answering questions or generating text. They’re designed to plan, reason through problems, and take sequences of actions. Think of them as models built to do work, not just respond.
Which Tools Use Which Models?
This is where it all comes together. A lot of business leaders hear about AI tools and assume they’re all more or less the same thing. They’re not. Here’s a quick look at how some of the tools you’ve probably heard of map to the types of AI we just covered.
| Tool | Type of AI | What It’s Mainly Good For |
|---|---|---|
| ChatGPT (OpenAI) | Generative AI / LLM | Writing, Q&A, brainstorming, code |
| Claude (Anthropic) | Generative AI / LLM | Writing, coding, analysis, and reasoning. In my experience, noticeably stronger than ChatGPT in coding and longer-form writing tasks |
| Microsoft Copilot | Generative AI / LLM | Productivity tasks across Microsoft products |
| Google Gemini | Generative AI / LLM | Writing, research, integration with Google tools |
| Salesforce Einstein | Machine Learning | Sales forecasting, lead scoring, CRM insights |
| Amazon Personalize | Machine Learning | Customer recommendations, demand forecasting |
| AutoGPT / AgentGPT | Agentic AI | Automating multi-step tasks with minimal input |
| Custom-built agents | Agentic AI | Workflows specific to your business |
The takeaway here is pretty simple. When someone says “just use ChatGPT for everything,” they’re talking about one type of AI built on one type of model. It’s a capable tool, but it’s not designed to do everything, and it’s not always the best option even for the things it can do. I use multiple tools depending on what the job calls for, and I’d encourage you to think the same way. It’s like using a word processor to manage your inventory. Sure, you could make it work, but there are better options.
Custom-built agents and fine-tuned models exist specifically for situations where a general-purpose tool isn’t enough. We’ll talk about when and why those matter in Part 2.
The Bottom Line
AI is not one thing. It’s a whole landscape of technologies, and the type of AI that makes sense for your business depends on what you’re actually trying to accomplish. Machine Learning is great for predictions and pattern recognition. Generative AI is great for creating content and working with text. Agentic AI is great for automating complex workflows.
Knowing the difference is the first step. The second step is figuring out how to put it to use, and that’s what we’ll cover next time.
If you want to get Part 2 as soon as it drops, make sure you’re subscribed to our updates. And if you’ve already got questions about where AI fits in your business, we’re happy to talk. No pressure, no pitch, just a conversation.
Great Lakes AI Solutions helps businesses across the Midwest build AI strategies that actually make sense for their operations. If you want to explore what that looks like for your company, get in touch.
