the term “slop” has been so successful as a way of dismissing AI content with the appropriate level of care (none, as nobody who cared made the slop) that I’m starting to see AI boosters try to change the meaning of the word. it’s important to push back when you see these information colonizers:
- insist that genuine human effort is slop because it isn’t to their taste (ie the new Mario movie)
- claim that their AI output isn’t slop because they “checked” it and found it to be particularly good
also, not that most people need the reminder, but AI boosters are out of their fucking minds and have no ability to evaluate quality.
you really don’t need to believe these conmen when they pretend a successful movie is on the same level as the trash they’re pushing.
you also don’t need to believe they’re checking their AI outputs in any way that matters. the point of LLMs and generative AI is to create spam. if they were into putting in effort, they’d do it without the AI.
I shouldn’t have to say this because this isn’t even a thread about the Mario movie but uhhhh
you’re allowed to not like it. I liked it, but there’s valid reasons to not like the movie on its own merits, to refuse to see it because Nintendo sues its own fans into lifelong debt, or to avoid it due to Illumination’s previous work
making up a version of the movie that doesn’t exist and then declaring it’s slop because that fake version sucks is fucking wild and I will make fun of you
also, even bad art is still art as long as it’s completed with human effort and care
like especially in movies, there’s so many awful movies that came out and had zero impact and got rediscovered decades later and enjoyed as outsider art
there’s movies that came out and were fantastic but got terrible reviews, and now they’re considered some of the best movies ever released because genuine art can be re-evaluated later (see also The Thing, which bombed so hard Carpenter almost quit films)
the Mario movie isn’t high art, it’s not particularly deep
and honestly thank fuck for that. not all art needs to push boundaries. sometimes you paint a really fucking good landscape and somebody hangs it in front of their toilet to stare at while they take a dump. most art ends up like that. it’s still art.
@zzt I feel like one thing it did particularly well was to incorporate scenes where they did exactly the thing they did in the game, without making it feel super awkward
I think the Sonic movies are way better tho
@jordan I really liked how the music built on the music from the games fairly naturally, and there were a bunch of minor gags throughout the film that were really well done and elevated the scenes they were in.
the sonic movies are surprisingly excellent! they added so much depth to the characters that the movie versions are the ones I think of first now. and I went into the first one prepared to hate it. I was so confused when like 15 minutes in I was like “wait I like this? it’s good????”
@zzt on the last point: I think most fail to understand the difference between use value and exchange value. The use value of LLM output can be high, but the exchange value is practically zero since anyone with model access can reproduce it.
This is also why I eschew the notion of successful enterprises built on LLM output: the fewer the people, the less effort involved to clone, the less value.
@zzt Ah.
1) AI creates slop.
2) Anyone who has worked with AI and found the results were not particularly sloppish, is obviously just an AI activist. Because, see 1)
Isn't that the kind of logic a sloppy AI would generate?
@zzt I'm enjoying pro Ai blogs like Mike Masnick's Techdirt desperately trying to ignore genuine concerns about AI and the people that are pushing it and glomming on to anything and everything they can find to try and paint everyone with concerns as medieval witch Hunters or McCarthyists because some people online accused something of being AI slop incorrectly, then getting yelled at for a hundred comments because they're so far out of line with their own readers.
You can tell they're flailing when the defense of the plagiarism engine requires smears or pushing groups who are opposed on every other issue together to create hypocrisy.
Perhaps the best words to describe the new Mario movie would be "sloppy", "slopish" or "slop-a-like".
The Guardian review of the movie says it was obviously carefully created so it can be dubbed into many languages as possible. Its plot is basically an average plot for a hero rescuing the heroine story. The reviewer said they used the word average to mean "mathematical mean" not "acceptable". Again this allows it to appeal to the widest possible global audience.
I suppose we should be grateful that a committee of people working for a film studio actually sat down and devised all of this rather than just writing a prompt for an "AI" and sitting back to collect the profits. At least humans did _something_ after all.
Its a sad state of affairs when people are working hard to deliberately create "content" which closely resembles slop. The use of the word "content" is another massive problem of course. How about we go back to using "art" instead?
@robcornelius > Its plot is basically an average plot for a hero rescuing the heroine story.
are you familiar with “Mario”, of the video game “Super Mario Bros” and also did you actually watch the same movie I did
@jackemled @zzt I have to disagree. A large language model doesn't understand what it's output means. It only understands that the words are statistically likely to occur in human created literature on similar topics to the prompt. The "AI" overlords tell us this the same process as how humans create art so the output of their machines is somehow art too.
The "AI" content is the lowest common denominator and statistical tricks pretending to be something it is obviously not.
I think when a humans try to make a piece of art that appeals to the largest number of people across as many cultures, demographics etc as possible they are emulating the "AI" / LMs methods. Whatever is created is stripped of all context, cultural significance and art.
Good art is different and even shocking. It's novel, not a remix of what has already been done time and time again to the point where it is meaningless.
If the whole "appeal to the widest range of people" applied in cuisine then the rancid, rotting juice in the bottom of a dumpster outside a ghost kitchen preparing delivery meals for a dozen or more restaurants with wildly different cuisines would be the food with the broadest possible appeal. That or vanilla ice-cream
@zzt checking AI output may be the difference between negligent slop and slop, but it's still slop.
For the argument's sake, I would concede someone, that their AI generated content is not slop, if they didn't only check it, but considered a significant amount of alternative outputs. E. g. if they generated 100 sentences and checked all of them and chose one. But that's obviously not feasible.
Infinite monkeys will at some point type Hamlet, but to be Shakespeare one needs to recognise Hamlet.
Tina
[object Object]
jordan
Edvin Malinovskis
Alex, the Hearth Fire
Torbjörn Norinder
Papageier
David Gerard
Rycochet
Matthew Cowen
Arthfach 🐻
Monokeros
robcornelius
Luna Lactea
Konstantin Weddige
Blake C. Stacey