26/08/2026
What exactly is AI? What exactly are LLMs?
I answered this question a few days ago, and in doing so, it occurred to me there may be a lot of people out there who are familiar with Artificial Intelligence and Large Language Models but don’t really know what they are.
So I thought I’d share a revised version of my answer.
The short answer is that humans have learned how to program computers to replicate aspects of human thought in a manner that allows computers to do a lot of things that used to require a human brain.
We call it artificial intelligence because, well, it ain’t human.
AI, at a base level, is software programmed to learn from data and make decisions.
LLMs are a specific iteration of AI, and companies like Anthropic (Claude) and OpenAI (ChatGPT) have created some of the most popular large language models currently in use.
LLMs are essentially pattern matching systems. They’ve been trained to see patterns in text by being fed billions of examples of language, and their job is to predict what word is most likely to come next.
Because these companies are training their software on massive amounts of text and data, they’ve created quite a few moral and ethical issues. Unfortunately, the people who created the data being used to train models aren’t necessarily receiving attribution or credit.
Until recently, attribution and credit for work you’ve created was considered a holy grail. It’s how a LOT of people create their living.
You created it? You get credit for it. You get paid for it!
If you take a computer that’s been coded to do this and then give it enough data, it starts to “think” and “act” in ways that resemble a human because much of the data it’s learning from was generated by humans.
Because the coding is human and the information it’s being fed has largely been created by humans (although that’s changing quickly because now most coding and some training data is computer-generated too), the computer’s output approximates human behavior and responses.
The more data it takes in, the better it can predict language patterns so, the “smarter” it gets.
What feels really weird is that now we’re seeing computer generated stuff right next to things created by actual humans and sometimes it’s getting really hard to tell the difference.
Obviously, there are still many things only a human brain can accomplish.
But we’re in a period where we’re actively learning what those things are because AI has evolved enough to approximate things like empathy, which until recently we thought was uniquely human.
Clearly, genuine empathy is still very human, but AI’s approximation is sometimes closer to healthy than, say, the empathy shown by an emotionally immature human.
We’re learning in real time that capabilities we used to assume required human consciousness or emotion don’t actually require those things, and it’s forcing the world to redefine what role computers should play.
And really, it’s making all of us ask ourselves, whether we like it or not, what it actually means to be human at all.