Earlier quoted context omitted.
If it can't yet, it will very soon. In less than five years probably.
And what is justifying such belief?
reality as we know it
How much of HN is AI?
241–250 of 370 posts
Re: How much of HN is AI?
#242This is categorizing with a pretty broad brush, it seems. I took a look at a random day, and the articles that are claimed to be AI are Waymo publishing a model for self driving cars and a post about scams in Apple News ads. Sure, Waymo’s model is AI and the scams in Apple News are often AI generated, but this isn't some sort of monoculture of LLM articles. Ironically, this article could have been improved by using a…
The reality is bots and bot-like behavior (honestly, much more likely the latter than the former from what I've heard IRL) are both equally effective in fucking up discourse on HN. It absolutely has gone to shit. And a number of longtime readers and commenters have called this out as well. The degradation started during COVID with the culture war crap (a plurality of accounts I bump into that comment and post are now…
plot twist seizethecheese is a bot themselves :D jkjk
Re: How much of HN is AI?
#243Re: How much of HN is AI?
#244Earlier quoted context omitted.
You really have to expand the list to ridiculous levels before you get to something as big. We've already left humans behind in many ways (but in others we've failed to build LLMs that know how to find the gaps in implementation that humans are leaving out) Cynics really say stuff like "but it's been 3 years and all they can claim is that it's changed text to speech, speech to text, photos to text, text to photos, pr…
> You really have to expand the list to ridiculous levels before you get to something as big Instead of expanding the list, it would be better to widen your vision. Email was more transformative than AI is right now. Unix, then Linux was more transformative. If you have a magic wand and remove AI, I think barely anyone would be impacted, but removing something like windows would be like erasing computers. And removin…
Email in it's first five years barely extended beyond some research universities. And is it novel? Do you see email as somehow a difficult concept to imagine?
Unix in it's first five years reworked the way operating systems performed. I'd agree it's pretty important, but I don't see how it fundamentally changed the way we perform work in multiple disciplines (in its first 5 years). Same for Linux, but even less so.
We're in very early times, and generative AI is touching every corner of the workpace and education, for better and worse. Powerful technologies can have negative effects. Email has been negatively transformative in ways as well.
Part of the reason you can erase AI so effectively with so "little" effect is that it is new.
Re: How much of HN is AI?
#245Earlier quoted context omitted.
The reality is bots and bot-like behavior (honestly, much more likely the latter than the former from what I've heard IRL) are both equally effective in fucking up discourse on HN. It absolutely has gone to shit. And a number of longtime readers and commenters have called this out as well. The degradation started during COVID with the culture war crap (a plurality of accounts I bump into that comment and post are now…
plot twist seizethecheese is a bot themselves :D jkjk
If they are so be it. Knuckedraggers and bots act the same way. At with some bots are kinda creative.
Re: How much of HN is AI?
#246This is categorizing with a pretty broad brush, it seems. I took a look at a random day, and the articles that are claimed to be AI are Waymo publishing a model for self driving cars and a post about scams in Apple News ads. Sure, Waymo’s model is AI and the scams in Apple News are often AI generated, but this isn't some sort of monoculture of LLM articles. Ironically, this article could have been improved by using a…
The reality is bots and bot-like behavior (honestly, much more likely the latter than the former from what I've heard IRL) are both equally effective in fucking up discourse on HN. It absolutely has gone to shit. And a number of longtime readers and commenters have called this out as well. The degradation started during COVID with the culture war crap (a plurality of accounts I bump into that comment and post are now…
I’ve always 1 indexed because there are non software engineers on here and I didn’t need to introduce the confusion that shibboleth would cause
Re: How much of HN is AI?
#247Someone made an HN without AI filter and it front paged here a bit ago. https://elijahpotter.dev/hnsansai (That's the first one I found, maybe there were others) So...are you complainers all using it? If not, why not? Spoiler -- because when someone says "I'm tired of reading about {X}", they're really saying "I want everyone else to stop talking about {X}". This has happened repeatedly over HN's life, and it just be…
A lot of us moved over to lobste.rs and it's been refreshing, to say the least.
Re: How much of HN is AI?
#248My reaction is: that's an appropriate amount. AI is probably the biggest thing to happen (in tech) in the last two decades, so I think it's fair that it's taking up a lot of the discourse.
Upvoted: I was going to write this exact reply. Everyone is a bit tired of hearing about AI, but it's advancing fast and it's the most important thing right now. As Demis says, if we get to ASI, we can solve all other issues with it.
Shit we’re onto thinking ASI is happening soon? I guess people hyped up AGI so much that it’s no longer thrilling to talk about despite not getting to that point either.
Re: How much of HN is AI?
#249Earlier quoted context omitted.
Pangram links several third party evaluations [1] that all seem to agree with 90+% detection, and very close to 0% false positives. Obviously those could be cherry picked, but I’ve yet to see arguments to the contrary that actually include any supporting data. (E.g here’s a prompt that will get Claude to spit out text that pangram doesn’t detect or here’s an article authored in 2018 that pangram says is AI.) Detectio…
> Obviously those could be cherry picked… The "could" isn't required here, since Pangram calls many of the evals "collaborations", and some received direct support. Results that don't favor Pangram as being among the least-worst of a questionable product category* surely wouldn't be featured. * https://mitsloanedtech.mit.edu/ai/teach/ai-detectors-dont-wo...
The most recent reference in that article appears to be from 2023.
Re: How much of HN is AI?
#250Earlier quoted context omitted.
I really wish there was a way to filter out discussions about AI on HN. It feels like lately they take up a good 1/3 of posts on the front page. I don't want to hear about some new AI startup.
Having a set of tag presets would help somewhat for filtering.
HN has been strongly adverse to tags for quite some time - for legitimate reasons. I don't think we'll see it anytime soon.