logoalt Hacker News

zacmpsyesterday at 7:59 PM1 replyview on HN

LLM summaries of papers often make overly broad claims [1].

I don't think this is a good example personally.

[1] https://arxiv.org/abs/2504.00025


Replies

crazygringoyesterday at 8:38 PM

When there's nothing else to go on, it's still more useful than nothing.

The story was being upvoted and on the front page, but with no substantive comments, clearly because nobody understood what the significance of the paper was supposed to be.

I mean, HN comments are wrong all the time too. But if an LLM summary can at least start the conversation, I'm not really worried if its summary isn't 100% faithful.