Edit to add: I also found someone who recorded a voice chat of the same thing. This isn’t that someone uploaded a song, or that AI didn’t actually process the file. These models really are this sycophantic:
Love it when AI says “it feels like”.
I love that we are killing the earth for this
Yeah, very atmospheric no doubt
Well, if it were my farts it nailed the “bedroom/DIY texture” portion of the assessment.
Everyone knows that the only acceptable use of AI is makinh kung fu videos with cats.
It’s the same stuff if you ask it to help you master a foreign language in dialogue mode
-
Help me practicing English pronunciation"
-
Sure thing! Now say “travel”
-
*user just belches in the mic*
-
great job! You really aced the accent!
-
He lied, it was a single fart from him lasting 37 seconds. slow clap
The reality is that it is a next word prediction machine. There probably aren’t any examples in the training data where people are writing music reviews on something that isn’t music. It probably interprets the sound as best as it can as “music” (and the best it can do this is likely very bad in the first place), and then, since the prompt was about reviewing music, it uses next word prediction to write a music review, which of course turns out looking like a typical music review. It’s not really interpreting the sound as “not music” especially since you told it it IS music.
True, but the principle behind the post is the beauty here. When not using the API, it costs these companies an unsustainable amount of money to make their models listen to fart sounds. I don’t use any AI myself, but I support anyone who wants to abuse the flat monthly subscription to make a company burn through money so that a plagiarism model can praise fart sounds.
Its in mp4, usually great for music and video, millions might enjoy listening to music such as the notes contained within this file.
Should I connect you with a music publisher for a quote? Should I publish to YouTube to monetize rights way…sorry right away?
Oh fuck, this gives me the idea to (locally) train an ai on fart noises and tell it its a way of communication to learn
Can you actually do this or are you speaking out of your ass?
Id have to look into it, but i am speaking out of my arse ( nice pun ) xD
maybe you will uncover the secret language of your ass
imagine farting in morse code to communicate with your computer.
Train it via text.
Quiet city montage… lol.
I’m old enough to remember an April Fools Day bit Adult Swim once did where every show just had consistent, random fart sounds including serious anime like Ghost in the Shell. I’m certain this LLM would approve of such a work of (f)art.
I don’t know if I’ll ever find it again, but there’s a “restored” version of that GiTS episode up on the internet archive. Someone took a bluray copy of the video and audio, and mixed in the relevant fart sounds from the audio of a VHS recording of the adult swim edit that was tracked down.
Lol wow, talk about dedication
The year they drew moustaches randomly on everyone was my favorite.
Faye Valentine from Cowboy Bebop had an epic one!
Inserting random fart sounds into cartoons literally sounds like my dream job.
I wonder if they actually linked it to an algorithm that analyses the sound that is looking for certain patterns or something, and that is why you get the “atmosphere piece” thing.
According to their tech/marketing papers, it’s supposedly multi-modal, encoding audio to tokens.
Jesus, what a complety fucking useless waste of resources
yeah but that doesn’t mean anything, does it? I don’t think they just tokenize the raw audio, that wouldn’t make sense, right?
I mean, you could. Just encode 100ms chunks or whatever into tokens then push them through the same model. I’m pretty sure that’s what the claim to do (though with MoE/routing now, maybe).
“Bedroom” texture 👌
Bedroom “texture!?”
That wasn’t just a fart. :(
for scatmen…
“… but AI said my idea was really good!”
