> Suppose you want to be really good at predicting the next word. If you want to be really good, you have to understand what’s being said. That’s the only way.

> Suppose you want to be really good at predicting the next word. If you want to be really good, you have to understand what’s being said. That’s the only way. So by training something to be really good at predicting the next word, you’re actually forcing it to understand. Ok, so in this framing you'll just get stuck below the "really good" level because the training isn't based on understanding