>>509273968>Just like how people said computers could never learn to play Go? We're not far from a Move 37 moment with LLMs.Oh, you are such a fucking midwit. You don't even understand how neural networks got good at go and why this is not applicable to LLMs.
Let me enlighten you, you complete imbicile.
AlphaGo Zero got good by self play, the previous version which used the human player data was garbage.
Self-play is what enabled the exponential growth.
That shit just doesn't apply to LLM, all we do is feeding it more and more text expecting it to get better. It's just not going to happen.
You are a fucking retard. Leave the thread in shame and never talk again.