Hacker News Reader

Why would it? It just learns by making this kind of text more likely to appear. If you want it to be smart it needs to see problems and strategies to solve them, not see explanations.

Regurgitating explanations isn't useful, following problem solving patterns is. So to learn from this article it would need to see the kind of thinking required to write the article.

View on HN

Okay. Are humans different than this?

Thoughts -> Action -> Thoughts

Thoughts are generated by brain. You don't control how the neurons are fired in the brain. As far as you are concerened you experience the thoughts, the sounds, images etc which.

So what exactly is the point of this article?

When you read something you typically imagine a lot of examples that the thing you read could apply to. The way we train LLMs today doesn't do this, it just reads the text without thinking further about it. That means to make an LLM learn you have to feed it examples instead of descriptions, like examples of problems being solved etc.

Maybe we could make LLM training do such things in the future, but it doesn't do it today and it is hard to do that in practice since generating examples on the fly for descriptions isn't easy to do in an intelligent way. I think that is a core part of generalizing knowledge so probably one of the keys we need to get to AGI.

Reply on news.ycombinator.com