AI breakthroughs aren't new math, just massive scale. Transformer networks from the 1980s work the same way, but now trained on enormous datasets with billion-dollar computing power. These models don't understand language—they predict what words come next based on patterns.
So they can do more things, but they're not fundamentally new math, but they're fundamentally new computing capability. But mathematically, the current state GenAI models are basically transformer models, which are layers of neural networks. Neural networks are what we started playing with in machine learning in the '80s. So the math really hasn't changed that much. The amount of data and the computing horsepower and transformer models are clever. They can get weights beyond different layers of the neural net so they can They do more things, but they're not fundamentally new math, but they're fundamentally new computing capability. And with the amount of data that we have, you get brand new models. There's a reason why people are spending billions and billions of dollars on data centers now, because that's the thing that's different, is the ability to use brute force, these enormous neural networks, multidimensional, to be able to figure out if I have this language input, I'll predict what should be the output. Now, your basic GenAI large language model doesn't really know what it's saying. Some people call it a The difference is computing. The difference is data. The difference isn't mathematical theory. That's pretty much stayed the same or evolved incrementally. It's not evolved revolutionarily.
