logoalt Hacker News

petesergeanttoday at 12:53 PM0 repliesview on HN

> the empirical track record of "LLMs can't..." is in itself a reason to call it into doubt ... Every "can't" of this nature was followed by a discovery of "they can, just poorly", and then etc

What LLMs are "fundamentally incapable" of doing has striking parallels to https://en.wikipedia.org/wiki/God_of_the_gaps