[bbor]

He wrote an op ed recently with the oft-repeated argument that stochastic models are fundamentally distinct from structured symbolic/logical models. In simple terms you can never really “trust” chatgpt’s answers because it’s just guessing the answer that looks right, not applying structured reasoning like humans do.

https://www.nytimes.com/2023/03/08/opinion/noam-chomsky-chat...

But might be missing something more specific, would love to read that.

[Izkata]

> In simple terms you can never really “trust” chatgpt’s answers because it’s just guessing the answer that looks right, not applying structured reasoning like humans do.

Here's an argument otherwise, that they really are developing some sort of model: https://the-decoder.com/stochastic-parrot-or-world-model-how...