> LLMs are nothing close to AGI and not going to lead to it, they can’t distinguish right from wrong, they can’t count, they can’t reason, they generate plausib

> LLMs are nothing close to AGI and not going to lead to it, they can’t distinguish right from wrong, they can’t count, they can’t reason, they generate plausible text from a vast databank of connected text. Argument? Are LLMs close to being able to significantly help AGI researchers?