What is ChatGPT and LLMs?
Okay, so by now you already understand that a machine made to do something extremely exact is an automation, not an AI.
You also already understand that AI is just a way we decided to organize computer components, imitating the human brain, and this started generating results very similar to human thought.
To the point where we can assert: AI thinks.
It doesn’t have feelings, it doesn’t have consciousness, but it thinks.
In a way different from humans, with similarities and differences, but it thinks, because it’s an artificial brain.
And you understood that it also learns and that it also makes mistakes.
Given all this, what are ChatGPT and these so-called LLMs?
Well, technically, LLMs stand for “Large Language Models,” which are systems trained on an absurd amount of data to understand and generate human language.
But, getting back to practicality and simplicity, think of them as types of brains.
So ChatGPT is a type of brain.
Actually, they are several ChatGPTs: there is ChatGPT 4, ChatGPT 3.5, and so on.
Each one is a type of brain.
And when you’re going to build your AI agent, you literally choose the brain you’re going to put in it, along with the training you’re going to give it, and other characteristics.
And this choice of brain is very important.
There are brains better for math, others better for creating software, others better for texts, others for conversations…
There are even generalist brains that are good at several things.
And there are brains that are worse at everything, that you can even think of as “dumb brains.”
But hold on, don’t write them off yet!
You’re going to see that even these dumb brains have their role when we talk about managing AI and building teams of AI.
But that’s a topic for later.
For now, what you need to understand is that when you go to the ChatGPT website, you are entering a chat with that specific brain of ChatGPT, with that LLM.
But when we talk about managing AI agents, we stop going to the ChatGPT website and we take its brain, and other brains, to build our own agents.
That’s an important difference that we’ll explore in depth in the next chapters.