ChatGPT's math is the same as logistic regression from decades ago. What changed: massive training data, cheap computing power, and a simple interface that let anyone use it. The UI is what made AI suddenly feel new.
What's different about why all of a sudden ChatGPT showed up? The interface. The interface is how you interact with everyone else. Neural networks are mathematically isomorphic to linear regression models. They're the same underlying math. And so that thing I was talking about, logistic regression and multivariate, that's the same math all the way through. What's different? The amount of data and the amount of horsepower to be able to crunch the numbers that we have today.
And a UI. I mean, I think that's part of it. You know, they talk about ChatGPT. They gave me something that I can interact with and get results. I don't go to a data scientist and ask.
That's the key. And I think what's different about why all of a sudden ChatGPT showed up— is it that fundamentally different than earlier technologies under the covers? No. But the interface, the interface is how you interact with everyone else. And so all of a sudden you have this very compelling way of interacting with a system that you're very comfortable with.
