notdevin

Gary Marcus argues that LLMs aren’t as good as a calculator based on their inconsistencies. I’m not convinced the tool Gary is thinking about is the same tool as what an LLM is useful for. It could be true that his ideal model is absolutely better but based on the way he’s so afraid of things given the words he uses, I doubt it

Gary Marcus argues that LLMs aren’t as good as a calculator based on their inconsistencies. I’m not convinced the tool Gary is thinking about is the same tool as what an LLM is useful for. It could be true that his ideal model is absolutely better but based on the way he’s so afraid of things given the words he uses, I doubt it

https://open.spotify.com/episode/7DGxH45T1S6iuVMdS88D1k?si=ZYMk-3XiRfmas9BxcSvVbQ&t=756&context=spotify%3Aplaylist%3A37i9dQZF1FgnTBfUlzkeKt

Catch me on X, notifications are off here 
@devinjelliot

I’m getting more convinced, the further I get, that Gary Marcus doesn’t understand humans as well as he thinks he does

The more I listen to him, the more I actually just hear a cogent argument about usefully ways to think about building ai companies. He’s just wrapped his positive argument for ai in a negative framework. If he took that perspective then he’d come off a lot better

I’m lost.. I did a little research and since LLMs are basically AI, what exactly is his argument?

I do think he is the most genuine of the ai risk people and seemingly not that crazy in how he cares about it

Medical student || music lover || Artist ||nature lover|| Food Lover. 🌸

Oh I thought the inconsistencies could be resolved .. didn’t know the usefulness of the model relies on it inconsistencies

I don’t think you can sort out the inconsistencies and still have a useful model. How do you mean sort out here?

What if the inconsistencies with LLMs are sorted out, I doubt there will be any comparison whatsoever

Comparing an LLM to a calculator oversimplifies their function. LLMs aren’t designed for perfect consistency like calculators; they excel at understanding context, language generation, and solving problems where ambiguity and nuance exist.