Not proof reading what you copy and paste really sells the shit quality detail you'll have as a business.
post
Did they copy paste that in and not read any of it? Normally I'd say that's a sign it's fake but with the slopbrains it's plausible.
And if that wasn't enough, nothing says "my readers are morons and I hate them" more than using the most generic default AI face for the "writer."
And the account has done the same thing on countless posts.
The fact they specify "in english" at the end makes me think they don't speak english and are just copying and pasting the prompt.
But like, sure they can estimate the different between a few dozen and two thousand?
That would be if they are vaguely paying attention.
Adhering to target word/character counts is not something the models are terribly great at, and if they were interacting in Chinese, they would have seen the result looking longer and by Chinese script standards, that size of words could look consistent with how much 2,000 words in mandarin would roughly look.
Having worked on a project that required English to Chinese translation with specific target character counts, I at least understand how easy it would be to have no idea and not even be able to rough eyeball estimate things.
Write me a 30 words comment on this post. Make no mistakes.
Write a snarky reply to the top comment that establishes they are wrong and I am the superior lemming. Try to maximize upvotes and don't make any tyopa or people will make fun of me.
There are also people trying to sell fake crochet stuff or botched patterns. I’m so thankful I’m experienced enough to know what works and how to tweak patterns, but newbies will be duped, get frustrated that the pattern is not turning out right, and possibly give up before improving. I bought some physical books published pre-AI just to have something else to reference.
I think at this point, AI systems are either manufacturing themselves or people are helping to generate them as quickly as possible for some money making scheme ... and all of them are doing it all so much and so fast that there is bound to be blatant appearances like this. For every slip of the keyboard, there are probably a hundred more instances where it came out as clean AI slop that no one noticed.
We're well on our way to a dead internet .. partly because AI is filling it with nonsense and partly because we are actively helping to make it happen.
This one is straight up manual copy-paste from a chatbot interface gone wrong. Not an automated slop farm.
But it would be trivial to create an automated slop farm that wouldn't run into this issue. Just need a cron job and some rudimentary scripts to call a few API endpoints. A regular generative model and something to classify the quality of the copy.
No need to have AI systems manufacture themselves, just set up an automated slop farm in 20 minutes and forget about it.
online recipe sites have been slop for over a decade, AI blogs have shifted it into overdrive
Seen that sort of thing on websites a few times now. Even without it being so blatant, so many websites seem AI generated these days.
I do like how the former anthrpomorphization of LLMs has given way to verbally and emotionally abusing the chatbots. Smth about that is funny to me, even though the whole loss of originality itself is rather bleak

all 28 comments