ChatGPT: Gary Marcus & Noam Chomsky are STILL in full delusion-mode
Gary Marcus (NYU), Noam Chomsky (MIT) and friends are professional GPT naysayers still hard at work convincing the world that OpenAI’s LLMs (large language models) as a whole are no good at NLU (natural language understanding), have nothing to contribute to linguistics and that AI is — hear this — not about artificial cognition.
All evidence to the contrary. It’s hard to keep a straight face.
Their latest claim is that ChatGPT & GPT aren’t a threat to Google. (Wrong! . . as we shall see. But Google is on it of course).
All I’m trying to do is present a balanced point of view on LLMs vs symbolic AI, for which I’m a practitioner and fan of both. I literally have no axe to grind.
(BTW: here’s a recent update post of mine on this issue).
Quick recap
A little over a month ago I wrote about what I am convinced is no less than gaslighting of the AI community by these researchers.
At best they are exaggerating hugely that ChatGPT ‘can’t do understanding’ (yet top scores on any metric you throw at it).
At worst (and this is a definite claim they make) they are trying to tell us that AI literally isn’t about cognitive software. That means they are…
