Discussion
RE: https://fosstodon.org/@libreoffice/117518007825825230
People keep misunderstanding the term slop.
It's not a definition of perceived quality.
It's a definition of process.
Slop can be "high-quality" and it's still slop.
People who produce slop don't get out of deserved ridicule by saying "well this isn't slop, because it's good."
@jaredwhite Agreed. Whether it's gruel or Bolognese sauce, it's still slop.
@jaredwhite whether the flash-flood is sewage or gold and jewels, it'll still knock you down and sweep you away
@jaredwhite this argument again?! 😆
For my money the defining quality of "slop" is that it's low effort. Listicles of the 2010s on Buzzfeed et al? Thats slop, just not AI Slop
@jaredwhite Instead of improving software for end users, let's focus on ideological personal opinions on what tools are people supposed to use. And if they use the wrong tools, let's exclude them from the community and participation. We don't want that kind of inclusivity. /s
@Chapz You have it exactly backwards. LLMs foster *exclusivity*, erode trust, corrode craft, and ultimately are a net negative for FOSS. Also agentic coding services are not just "tools" any more than "Soylent Green" was just "food."
@jaredwhite I disagree, because all I see all over the internet are only extremes. One side says AI is the ultimate good and can do nothing bad, and the other side says AI is ultimate evil, and unable to do any good.
The truth is but in the middle. AI is like a knife, it can be used either to prepare food or kill people. But it's up to the human using it which one is it going to be.
@jaredwhite I agree there is slop, by the original meaning of the word, low quality code which brings no value and the owner doesn't even understand it. But, can you agree, someone with experience and understanding of the code, which doesn't use automatic agents, but guides and controls it's every step, can produce amazing things?
@Chapz I think you're speaking in the hypothetical.
The original version of GitHub Copilot was the "spicy autocomplete" idea of just filling in a line or a few lines. I didn't use it even then, because the copyright situation was absurd. I figured MS would fix all that, improve quality by *narrowing* scope, and then *maybe* it would be a decent tool.
Instead, we ended up with the world of Claude Code et al. Prompting "build me a thing" and then getting a huge PR is not autocomplete. It's slop.
@jaredwhite I agree on that, but it's not the only use case. I understand my codebase, I consult with AI on architecture, possible directions to go, upsides and downsides, let it propose a solution, think about it, and then write parts of the code myself and let it write some of the boilerplate which I know exactly how I want it to be. It never does any single task, where I didn't give it precise details and scope what to do and how.
Is the result of this.. slop?
@Chapz I think you raise a good and nuanced point.
My own position is, actually, I mostly agree with you! I've been saying that I don't like to get on people's case when they use genAI for fuzzy search/brainstorming/prototyping/etc. I don't personally dig it, but it's not my primary concern.
I'm trying to keep stuff out of *production* assets. The artifacts the public directly consume in the form of executing code, prose, images, video, etc. *That's* where I'm drawing a line.
@jaredwhite I understand this as my example would not be welcome in production code, simply by the fact that AI was involved, no matter my expertise, understanding, control I exercised over it, or anything else?
I understand the line, but not the requirements to land on one side or the other by your standards.
@jaredwhite Blood diamonds are still diamonds.
@jaredwhite well, isn't it you who just redefined slop meaning "total garbage produced by LLM" with "everything produced by an LLM"? You may want to edit the wikipedia article to match your definition: https://en.wikipedia.org/wiki/AI_slop
@assimilateborg Merriam-Webster:
> We define slop as “digital content of low quality that is produced usually in quantity by means of artificial intelligence.”
Well, who gets to determine it's low quality is the question, isn't it?
Certainly not the purveyor of the slop. They'll always claim it's high quality. 😂
I say, they switched from ridicule "left behind" rhetoric, to fighting us by trying to gain Interpretive Authority over our words.
I pledge to ridicule the sloplords even more now. Maybe something good will come of it.
@jaredwhite slop can be also produced by human beings, as the text above proves.
Install Fedizen
Get the full app experience
@jaredwhite I think that’s part of the reason I’m not fond of the term. ‘Slop’ is too easy to read as a quality judgement, I wish we had a better term that called out the ethical (and other problematic things). It’s never ending for one to try to argue their output isn’t slop, but it would be harder to argue it’s not touched/tainted (for lack of a better word) by LLMs.
(Languages and clear communication are hard ^^;)