commented:
This is the same root complaint I have with those dumb things that
generate slop-wikis of my open source work and docs.
It's exactly the wrong use. It's an assault on the author and on the
creative community.
It replaces value-positive communication with value-negative bullshit

  commented:
Totally; one of the reasons I've switched to paying for Kagi search is
that they filter out sites like deepwiki that get in the way of me
finding interesting content. It's easy enough for me to generate my
own LLM summaries if I feel I need it, based on my own prompts. It's
much more difficult to locate other interesting human perspectives on
the Internet while wading through slop.

  commented:
This seems like a great example of cluelessly saying "let's throw AI
on it."
One problem is that the AI doesn't seem to be very good. A second
problem is that this seems like a really dodgy idea about how to use
AI. As the post notes, authors already write abstracts. Unless the
author has done a bad job, the value of having AI redo it is pretty
low, even if the AI were good.

  commented:
Glanced through a summary myself and it gets enough subtleties wrong
to be possibly misleading; and yes, I agree with the author that in
theory a paper's introduction should just be a better version of this.
Although let's be honest, there are many papers out there with shall
we say sub-optimal writing.
Ignoring confabulations (because in my case there weren't all too many
important ones at a glance), I still took issue with the summary.
While I read both it and the OP, I found myself reminded of The Monad
Tutorial Fallacy, specifically this part:

.... he finally has an “aha!” moment: everything is suddenly clear, and
Joe Understands Monads! What has really happened, of course, is that
Joe’s brain has fit all the details together into a higher-level
abstraction, a metaphor which Joe can use to get an intuitive grasp of
monads ... The problem, of course, is that if Joe HAD thought of this
before, it wouldn’t have helped: the week of struggling through
details was a necessary and integral part of forming Joe’s ...
intuition, not a sad consequence of his failure to hit upon the idea
sooner.

I don't believe that papers are perfectly minimal ways of conveying
ideas: certainly there is (perhaps lots of) room to summarize. But I
do think that to be truly useful, a summary needs to align with
whatever high-level abstractions work for our own brains, and the
uncanny feeling I got from reading the AI summary was a high-level
abstraction that was out of alignment with own way of thinking in
several places.
What I fear is that LLM summaries are increasingly being used as
shortcuts to intuition building which either requires (1) puzzling
through less than "perfect" explanations or (2) finding an explanation
"perfect" for your brain. The danger I see is the presentation is
meant to always give the appearance of (2), even if it is off (again
--- ignoring confabulations), thereby building bad intuitions like
"monads are just burritos."

  commented:
ACM is about to make papers free to read or download: yay, finally! A
sizable swathe of all computer science research will soon be available
without a subscription. Now, they need a way to fill the revenue gap.
They're a nonprofit doing useful things with money, I assume. I'm no
expert on the ACM's activities but I imagine we'd all miss them if
they were gone, so I support their attempt to make money from
something besides paywalling papers. Unfortunately, my first
impression of this "ACM premium" subscription is that it isn't
worthwhile for me and I doubt it will be for anyone else. But the harm
here is that the ACM has made an effort to fill their revenue hole and
it looks like the effort will be wasted and the hole will remain. The
harm is probably not that there will be some more bad AI summaries on
the web (the summaries will be paywalled and no one will pay to read
them) and the harm is not that they're paywalled (again, because no
one will want them, so restricting them harms no one).
Take all this with a grain of salt, I have no special knowledge about
the ACM, surely they've thought about this much more than me so I'm
probably wrong.
.