3 ms·

I actually disagree, in certain cases. Just today I saw: https://news.ycombinator.com/item?id=46204895 https://news.ycombinator.com/item?id=46204895 when it h

I actually disagree, in certain cases. Just today I saw: https://news.ycombinator.com/item?id=46204895 https://news.ycombinator.com/item?id=46204895 when it had only two comments. One of them was the Gemini summary, which had already been massively downvoted. I couldn't make heads or tails of the paper posted, and probably neither could 99% of other HNers. I was extremely happy to see a short AI summary. I was on my phone and it's not easy to paste a PDF into an LLM. When something highly technical is posted to HN that most people don't have the background to interpret, a summary can be extremely valuable, and almost nobody is posting human-written summaries together with their links. If I ask someone a question in the comments, yes it seems rude for someone to paste back an LLM answer. But for something dense and technical, an LLM summary of the post can be extremely helpful. Often just as helpful as the https://archive.today https://archive.today... links that are frequently the top comment.

zacmps 11mo ago
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 https://arxiv.org/abs/2504.00025
- crazygringo 11mo ago
  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.
Rarebox 11mo ago
That's a pretty good example. The summary is actually useful, yet it still annoys me. But I'm not usually reading the comments to learn, it's just entertainment (=distraction). And similar to images or videos, I find human-created content more entertaining. One thing to make such posts more palatable could be if the poster added some contribution of their own. In particular, they could state whether the AI summary is accurate according to their understanding.
- crazygringo 11mo ago
  I definitely read the comments to learn. I love when there's a post about something I didn't know about, and I love when HN'ers can explain details that the post left confusing. If I'm looking for entertainment, HN is not exactly my first stop... :P

zacmps 11mo ago

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 https://arxiv.org/abs/2504.00025

crazygringo 11mo ago
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.

Rarebox 11mo ago

That's a pretty good example. The summary is actually useful, yet it still annoys me. But I'm not usually reading the comments to learn, it's just entertainment (=distraction). And similar to images or videos, I find human-created content more entertaining. One thing to make such posts more palatable could be if the poster added some contribution of their own. In particular, they could state whether the AI summary is accurate according to their understanding.

crazygringo 11mo ago
I definitely read the comments to learn. I love when there's a post about something I didn't know about, and I love when HN'ers can explain details that the post left confusing. If I'm looking for entertainment, HN is not exactly my first stop... :P