Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Another Corner of the Internet Has Been Ruined

freeformatter.com

Post reply on HN
Print view View original

51–60 of 84 posts

Re: Another Corner of the Internet Has Been Ruined

#51

Post by Tiberium »

pixl97 wrote: post #44
Tiberium wrote: post #25

Earlier quoted context omitted.

Can you show some examples of Pangram being "hot worthless garbage"? This reputation is true for most other detectors, but Pangram spends considerable effort to have an extremely low false positive rate. I think you just used some of them in 2023 and now think that they can't be made better.

Language is a living thing. At one time saying 'cool' to mean something was neat wasn't a thing. Then someone used it that way and it spread throughout humanity (or least English speakers in the US). This means as people use AI and read AI we'll start speaking like AI just as much as AI speaks like us. ESL people tend to have it even worse as they commonly speak a more proper subset of English and get identified as A…

Okay, I have a test for you. It's a bit undocumented, but I'm pretty sure HN is using Pangram or a similar tool to auto flag AI written comments. Can you copy the entire text from the linked article and reply with it here, and see what happens to your reply, would it get flagged or not? I think this will show you what reality we live in.

Re: Another Corner of the Internet Has Been Ruined

#54

Post by Arainach »

Tiberium wrote: post #28
anigbrowl wrote: post #13

Earlier quoted context omitted.

You are a fool if you think you can detect AI reliably from such a short sample. It seems not to have occurred to you that some people can write with expression and wit, and AI emulates their style .

You are a fool if you really think that the linked text is human written. I interact with LLMs basically daily, it's incredibly easy to spot their default styles. And, honestly, you should trust Pangram over me.

"AI Writing Detectors" are absolutely full of false positives, as anyone working in education or related fields can attest.

Re: Another Corner of the Internet Has Been Ruined

#55

Post by hyperhello »

reactordev wrote: post #32
hyperhello wrote: post #24

Earlier quoted context omitted.

So I had to copy-paste my source code into a web site, submit, then copy-paste the text back out into my document? I'm not putting down the work the author did, but that workflow was always going to be completely obsolete at some point.

for one-shots and non-developers, it was useful. Simply because you found no utility in it doesn't mean there isn't one.

I didn't say I found no utility, it could have tons of utility, but technology changes over time. Perhaps the gentleman author could have produced a desktop Electron app or something similar running locally.

Re: Another Corner of the Internet Has Been Ruined

#56

Post by Tiberium »

Twiin wrote: post #39
Tiberium wrote: post #25

Earlier quoted context omitted.

Can you show some examples of Pangram being "hot worthless garbage"? This reputation is true for most other detectors, but Pangram spends considerable effort to have an extremely low false positive rate. I think you just used some of them in 2023 and now think that they can't be made better.

I put your initial comment into Pangram and it said "We believe that this entire text is AI." So.

... Come on now. This is obviously because I quoted the LLM written parts from the article. Remove the quoted parts and try again. Do people not understand what LLM detectors do?

And, to be fair, short texts are one of the areas where even Pangram struggles, with a longer text it would be able to identify it as a mix of human and AI written text. They clearly disclose that short text detection is less reliable.

Re: Another Corner of the Internet Has Been Ruined

#57

Post by Tiberium »

Arainach wrote: post #54
Tiberium wrote: post #28

Earlier quoted context omitted.

You are a fool if you really think that the linked text is human written. I interact with LLMs basically daily, it's incredibly easy to spot their default styles. And, honestly, you should trust Pangram over me.

"AI Writing Detectors" are absolutely full of false positives, as anyone working in education or related fields can attest.

Please see https://news.ycombinator.com/item?id=49212433 and try yourself. And Pangram has been confirmed to be working by reputable educational institution, too.

Re: Another Corner of the Internet Has Been Ruined

#58

Post by Tiberium »

Tiberium wrote: post #12
rpdillon wrote: post #11

Earlier quoted context omitted.

"It was written with AI!" is boring (and often wrong). It shows up on almost every post, and doesn't require any skill or insight. That's why GP said "Can you please not?" It's just not interesting.

I only post it when I'm sure and I cross verified. And in this case specifically it's not boring, quite the opposite, it just shows that humans can be quite inconsistent. I really hope HN would extend its no LLM rule to articles/text, although this is too hard to enforce. When I look at HN, I want to read human text, I don't mind the use of LLMs for actual code and use it a lot myself. My issue is with people who gen…

To anyone who's willing to do a short HN test, please see and try https://news.ycombinator.com/item?id=49212433

You're in for a nice surprise.

Re: Another Corner of the Internet Has Been Ruined

#59

Post by D13Fd »

aleksejs wrote: post #45

I'm not sure I understand the intended message here. The author seems to describe their own project as "cover[ed] ... with invasive advertising" while saying that new technology accomplishes the same outcomes "faster, cheaper, and [with less maintenance burden]". So what exactly has been ruined?

Exactly. And having AI to work out the 2am dependency issues actually makes maintenance a lot easier. This seems like an odd message.

Re: Another Corner of the Internet Has Been Ruined

#60

Post by pshirshov »

In this particular case - good riddance. Such tools should not require a round-trip to an ad-infested website on the other side of the world. The crucial issue is that in the process you have to trust the owner (and perhaps ad network owners) with your data. Even if you want a round trip, you could have a static page somewhere on GitHub. The fact that the models made this business model obsolete is not necessarily a bad thing.
Post reply on HN