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

Antislop: A framework for eliminating repetitive patterns in language models

arxiv.org

Post reply on HN
Print view View original

11–20 of 119 posts

Re: Antislop: A framework for eliminating repetitive patterns in language models

#11

Post by yawnxyz »

mrbungie wrote: post #3

I don't know if not getting the idea right, but I'm pretty sure people refer to AI outputs as "slop" not due to (only) repetitiveness. According to some sources: [1] Wikipedia > AI slop is digital content made with generative artificial intelligence, specifically when perceived to show a lack of effort, quality or deeper meaning, and an overwhelming volume of production.[1][4][5] Coined in the 2020s, the term has a p…

Honestly "slop" should also be retroactively applied to e.g. Buzzfeed content; it shouldn't just be AI-centric

Re: Antislop: A framework for eliminating repetitive patterns in language models

#13

Post by louthy »

yawnxyz wrote: post #11
mrbungie wrote: post #3

I don't know if not getting the idea right, but I'm pretty sure people refer to AI outputs as "slop" not due to (only) repetitiveness. According to some sources: [1] Wikipedia > AI slop is digital content made with generative artificial intelligence, specifically when perceived to show a lack of effort, quality or deeper meaning, and an overwhelming volume of production.[1][4][5] Coined in the 2020s, the term has a p…

Honestly "slop" should also be retroactively applied to e.g. Buzzfeed content; it shouldn't just be AI-centric

It isn’t AI centric, it’s derived from poor quality wet food. Often given to pigs or used to describe prison food. It’s the origin of the term ‘sloppy’.

Colloquially it means ‘poor quality’ and always has done. So buzzfeed is journalism slop, just like poor quality AI content is AI slop.

Re: Antislop: A framework for eliminating repetitive patterns in language models

#14

Post by moritzwarhier »

jsheard wrote: post #5
mrbungie wrote: post #3

I don't know if not getting the idea right, but I'm pretty sure people refer to AI outputs as "slop" not due to (only) repetitiveness. According to some sources: [1] Wikipedia > AI slop is digital content made with generative artificial intelligence, specifically when perceived to show a lack of effort, quality or deeper meaning, and an overwhelming volume of production.[1][4][5] Coined in the 2020s, the term has a p…

Yeah, what this actually achieves if anything is making it harder to quickly recognize slop for what it is, so readers are more likely to give it the benefit of the doubt and keep their eyeballs on it for longer. Which I suppose is desirable if you're in the slop-mongering business (e.g. doing SEO spam or other such methods of flooding the commons with sewage for the sake of profit).

Fits into a broad pattern of deceptive LLM terminology, for example "Deep Research": a humble and honest moniker would me "Reflection" or "Recursive self-prompting".

Re: Antislop: A framework for eliminating repetitive patterns in language models

#15

Post by atourgates »

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer.

Lots of things have changed in that year, but the things that haven't are:

* So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term).

* Random emojis.

* Affirmations at the start of messages. ("That's a great idea!") With a brief pause when 5 launched. But it's back and worse than ever now.

* Weird adjectives it gets stuck on like "deep experience".

* Randomly bolded words.

Honestly, it's kind of helpful because it makes it really easy to recognize content that people have copied and pasted out of ChatGPT. But apart from that, it's wild to me that a $500bn company hasn't managed to fix those persistent challenges over the course of a year.

Re: Antislop: A framework for eliminating repetitive patterns in language models

#16

Post by giancarlostoro »

atourgates wrote: post #15

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer. Lots of things have changed in that year, but the things that haven't are: * So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term). * Random emojis. * Affirmations at the start of messages. ("That's a great idea!") With a brief pause w…

I dont use ChatGPT very often, though perplexity has it, but I find that going all caps and sounding really angry helps them to fix things.

Re: Antislop: A framework for eliminating repetitive patterns in language models

#17

Post by estimator7292 »

atourgates wrote: post #15

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer. Lots of things have changed in that year, but the things that haven't are: * So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term). * Random emojis. * Affirmations at the start of messages. ("That's a great idea!") With a brief pause w…

Ah, you've hit a classic problem with :smile_with_sweat_drop:. Your intuition is right-- but let me clarify some subtleties...

Re: Antislop: A framework for eliminating repetitive patterns in language models

#18

Post by gowld »

atourgates wrote: post #15

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer. Lots of things have changed in that year, but the things that haven't are: * So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term). * Random emojis. * Affirmations at the start of messages. ("That's a great idea!") With a brief pause w…

> Honestly, it's kind of helpful because it makes it really easy to recognize content that people have copied and pasted out of ChatGPT

Maybe it's intentional, like the "shiny" tone applied to "photorealistic" images of real people.

Re: Antislop: A framework for eliminating repetitive patterns in language models

#19

Post by teeray »

atourgates wrote: post #15

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer. Lots of things have changed in that year, but the things that haven't are: * So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term). * Random emojis. * Affirmations at the start of messages. ("That's a great idea!") With a brief pause w…

You can take my em-dashes from my cold, dead hands—I use them all the time.

Re: Antislop: A framework for eliminating repetitive patterns in language models

#20

Post by antoniojtorres »

atourgates wrote: post #15

I've been using ChatGPT fairly regularly for about a year. Mostly as an editor/brainstorming-partner/copy-reviewer. Lots of things have changed in that year, but the things that haven't are: * So, so many em-dashes. All over the place. (I've tried various ways to get it to stop. None of them have worked long term). * Random emojis. * Affirmations at the start of messages. ("That's a great idea!") With a brief pause w…

The emoji thing is so bad. You can see it all over github docs and other long form docs. All section headers will have emojis and so on. Strange.
Post reply on HN