Media and search engines nowadays need a flag system and a filter for AI junk results
The same reason as why Spotify does it. Content that they don’t have to pay any royalty for to anyone. They can keep 100% of the ads money. In a few years we’ll find out all these accounts were just created by the platform themselves. Legally.
I mean, yes it is garage, but:
- This Popsie Funk channel is upfront, that the music is AI generated. (Link for convinience: https://www.youtube.com/@PopsieFunk )
- There are probably channels out there, that are not upfront and YouTube won’t be able to detect them.
- Why are you listening to the youtube recommendations? Why are you looking at the recommendations? What are you expecting there in the first place?
- Also you can train your recomendations pretty well, by opening videos you don’t like and give them a thumbs down.
Yes, it is still bad.
This Popsie Funk channel is upfront, that the music is AI generated.
goes looking
Yeah, the description reads:
Popsie Funk is a fictitious creation. The tracks are A.I. generated from lyrics and musical compositions that I have created. The A.I. samples are then mixed and edited by me.
I am adding this disclaimer due to repeated questions about the genuine authenticity of Popsie Funk and his music.
I don’t think that the artist in question is faking this.
All that being said, while this particular case isn’t, I suppose one could imagine such a “trying to pretend to be human” artist existing. That is, if you think about all the websites out there with AI-generated questions and answers that do try to appear human-generated, you gotta figure that someone is thinking about doing the same with musicians…and at mass scale, not manually doing one or two.
I am adding this disclaimer due to repeated questions about the genuine authenticity of Popsie Funk and his music.I don’t think that the artist in question is faking this.
All that being said, while this particular case isn’t, I suppose one could imagine such a “trying to pretend to be human” artist existing.
They were pretending, and added the disclaimer because people bugged them about it. They still worded the disclaimer as if Popsie Funk was a real person (his music).
Why are you listening to the youtube recommendations? Why are you looking at the recommendations? What are you expecting there in the first place?
There was a time when the YouTube recommendations were actually pretty good. I discovered a number of great artists just from suggested songs and playlists. That no longer seems to be the case though, my recommendations have been garbage for more than 5 years at this point.
Dunno what y’all did to your algorithms but mine is still working just fine. When something odd comes up that I don’t like, I tell it not to continue recommending that.
Yeah. You gota curate that shit. Then it works.
Well, I make maps for a rhythm game and like to rewatch my showcase videos, so it started offering me that (meaning it’s music I enjoy but with a lot of shooting in the background).
But other than that it’s solid.
It can definitely work if you’ve trained it right. It’s how I discovered many of my now favorite vocaloid/utau/deepvocal/whatever songs, amongst other groups outside of that vocal synth genre. It definitely isn’t doing me super dirty in the music department.
It could work when the algorithm was like that but they changed it to consider using users as test subjects for more lucrative users
Here’s the full text of the disclaimer on the channel’s “About” tab as of present:
“Disclaimer:
Popsie Funk is a fictitious creation. The tracks are A.I. generated from lyrics and musical compositions that I have created. The A.I. samples are then mixed and edited by me.
I am adding this disclaimer due to repeated questions about the genuine authenticity of Popsie Funk and his music.
While being asked the same question dozens of times can be taxing, I take confusion as a huge compliment!
After all, if you can’t tell by ear that my music is A.I. generated, then I’m doing my job right!”
The channel owner directly states that it is their intention to mislead. I did see the disclaimer on the channel after looking up the “artist” and before making this post, but that disclaimer is not visible on the thumbnail preview and the video description omits any reference to it. The inclusion of the year in the video title as well as the hashtags all attempt to work their way into the feeds of those not in the know to convince them that it is legitimate.
The channels that are not upfront are even worse.
When I am using my phone as opposed to a desktop, I watch YouTube videos in the phone’s built-in browser so I can refresh the page to skip any ads before the video. I typically don’t have the patience after watching the video to open the YouTube app and wait for an ad to load and then wait to swipe the ad out of the way just to “Like” or “Dislike” the video. I may glimpse through the recommended page on the chance there is anything that I may have missed, or that may have been a surprise upload, or that may be adjacent to videos/channels that I’ve already watched and which may be of interest to me.
The channel owner directly states that it is their intention to mislead. I did see the disclaimer on the channel after looking up the “artist” and before making this post, but that disclaimer is not visible on the thumbnail preview and the video description omits any reference to it. The inclusion of the year in the video title as well as the hashtags all attempt to work their way into the feeds of those not in the know to convince them that it is legitimate.
I agree in that regard.
The tracks are A.I. generated from lyrics and musical compositions that I have created. The A.I. samples are then mixed and edited by me.
Generated from human compositions, human-mixed, human-edited, there’s plenty of songs which have less human input. Even I can steal beats from a frying steak.
This isn’t the “automated AI slop” that you’re looking to complain about.
As to “intention to mislead”: That has nothing to do with AI. Passing off a new composition as a 1974 track on first sight is peak retro.
Indeed.
When popular “real artists” release songs, they usually won’t have written the lyrics or the music, leaving just the vocals, which they will auto tune and possibly even mime in future performances.
A producer will then use powerful software to mix and refine everything.
So really the question to me is not about “is there anything impure in this art?” It’s “where is the line?”.
by opening videos you don’t like and give them a thumbs down
You don’t need to do this. There literally is a feature for exactly this: Click on the three dots and on “Don’t recommend”. If you do this, content like that won’t be recommended to you anymore.
But then the content won’t receive my thumbdown :(
I can promise you that the ‘dont recommend’ rarely works, and if used on Youtube Music it will actually show it to you as if you liked the video, even if you disliked it on regular Youtube
Shit like this is why I don’t feel bad about my aggressive ad blocking on YouTube. If they don’t care to deprioritize garbage like this (or moderate really at all) then why should I sit through a 50 second ad to watch a 5 minute video?
So tangential storytime: my husband recently discovered one of these channels (not sure if it’s this one or just a similar one), and specifically this goddamned song called “I glued my balls to my butthole again.” He will. Not. Stop. Playing it. He thinks it’s fucking hilarious, especially the more I roll my eyes.
Anyway, yes these channels are a blight, something needs to be done about them, and if anyone needs a husband they can find mine on the curb blasting that song on a loop lmao
Lul I’m fairly certain obscurest vinyl isn’t one of those Ai channels. They’re actually pretty hilarious. I wouldn’t play them on repeat.But they’re fun for a laugh with friends who haven’t heard then yet.
That song in particular is a song I’ve heard of before, it was one of the first AI generated songs I found (and was sent around a bit because it was an early example of what ridiculous stuff one can do with AI).
You can also hear it to be honest, if you listen closely to the vocals. (Luckily you can still spot AI songs that way, I wonder how long that’ll work.)Also I looked up obscurest vinyl, and apparently all their songs are AI generated.
Well shit. Fair enough, I was misguided.
This is a statement from an AMA by the creator:
In a way, this sounds similar to someone using a DiffSinger or ENUNU voicebank to make music, but without having to write all the lyrics word/syllable for word/syllable on individual notes and change the notes they’re on individually because you pitched them wrong. Also without the backing track too.
Either way, interesting enough that you can even do this kinda stuff. Especially since any time I’ve played around with that kind of AI, it’s always never turned out even remotely halfway decent in any way.
Holy cope from that guy lmao
“uhh it’s still a significant effort guys, I might just suck at every part of making actual music but I work hard on pressing random buttons”
Closer to “I cannot play every instrument or sounds like older vinyl music stars, so I use AI to help replicate the sound and feeling of older music while writing new lyrics. I vet all work done by the computer and fine tune its outputs for the best results.”
You don’t need to play every instrument ever, samplers exist, free song samples you can mix exist, there’s so many resources and techniques you can use to replicate older music.
Guy is just shitting out awful ai bot spam on the internet but it’s ok because he gives it funny titles.
Username appropriate.
You will not see the kingdom of god
I love it! https://www.youtube.com/watch?v=wPlOYPGMRws
If you don’t like the titular hit 'i glued my balls to my butthole, again" , then fuck you
What do you think the word “titular” means?
He says the titular line in “I glued my balls to my butthole again”
boobs or something idk englinsh
Surfers say tubular, strippers say titular.
Hell yeah!
It’s honestly not as good as I Staple Tapeworms To My Penis.
I liked Putrid Mutant better
If you DO like them though, try 1-800 PAIN
Or " It’s Time To Take a Shit on the Company’s Dime", that one slaps as well!
Welcome to the world of tomorrow!
I glued my balls to my butthole again.
Looks like they disabled the comments lol
The technical term is AI slop.
I’ve been thinking a lot about the state of things, and where we are heading.
At what point do we consider the internet “useless”? It must be coming very soon (less than 2 years), since the majority of content will be AI generated and targeted, which drives down the value for users even further.
Once original ideas vanish, and you can’t trust any text/audio/video/photo you see, what will be the point? It’s like the internet will simply be a video game world with next to no value.
And I can’t see how society can possibly reverse this.
Saying the internet is going to be “useless” in less than two years might be one of the wildest takes I have ever read.
Let me ask you this: assuming you use the internet for information rather than entertainment, would the internet be useful if the majority of content ends up being AI generated (not fact checked, not accurate, and not original)?
What if the overwhelming content you come across could neither be verified as true, and the majority of comments (including here on Lemmy) were bots? Would you still use it?
For me, it would stop being useful. Almost like a library only carrying fiction, when I’m trying to research a topic.
For entertainment, sure, it’ll be great for sucking the attention from people without having to invest in skill to be good at something. Hell, if you currently find YouTube shorts and Tiktok to be “good content”, then it’ll be around forever. Corporations and advertisers love this technology.
Satire and parody are the only legitimate uses of AI, as those are explicitly transformative works. There’s another channel I like that makes songs that are actually funny.
Satire and parody are the only legitimate uses of AI
…text to speech, dictation apps, translation systems, enemies in video games, chess bots, cancer discovery, image, video and file compression, cluster algorithms, CGI, world generation, etc.?!
That statement is just ridiculous
Sorry I really did mean in published media
I agree with all of that. AI has it’s valid uses.
But the way we are seeing it being utilised is often simply to flood every corner of the internet with spam, bad information, low quality content, and loads of filler.
I’m personally amazed by what AI can do with photo generation, music creation, and other creative work.
But at the same time, I want to know that it’s AI generated and not passed off as human created content. Especially with written content.
AI-based tools can be amazing, but only if ethics are applied to their use.
if the majority of content ends up being AI generated
yeah…
The majority of the internet is porn. Literally. That doesn’t mean that every information you look up is porn. Just because there exist much of it doesn’t mean wikipedia and other credible sites will suddenly just disappear.
And honestly, while AI generated articles are absolute garbage, it’s not like LLMs you can chat with are completely useless.The majority of the internet is porn.
Again, I’ll separate entertainment from informational, since entertainment can be garbage, and still be consumed.
Bad information doesn’t help anyone.
it’s not like LLMs you can chat with are completely useless.
The problem is, you wouldn’t know unless you know.
With a legitimate website that has human writers, editors, and fact-checkers, they can at least have creditability and a reputation to uphold.
Far too many randomly generated websites have a lot of information, but without any guardrails. If you know enough about a topic, you’ll realise that the information on these AI sites are pretty much useless. That is, you couldn’t use them as a source because enough of the info is bad/incorrect/incoherent, that it’s like asking a toddler who may or may not give you a valid question.
I’ve contacted a manufacturer of bike stuff, and their support is given by AI. While the answers you get sound like they could be right, it’s like getting an answer from someone who heard something about something from a friend. When you actually ask for a human, the answer is often different (and correct).
There is no accountability, or credibility, or responsibility, or integrity with AI. It has no reputation to lose if the information it provides is bad or not.
I know that AI isn’t going away. I’d personally be OK with some human verification system for websites, and would be more than willing to use a filtered version of the internet that blocks AI generated content. Call it curated or whitelisted, but I want my information to come from a human being.
Conversely, I still need to make and receive important phone calls even though the vast majority of calls I receive are spam
But you know they are spam, so it’s something you can avoid. But what if the majority (over 80%) of the calls you receive can’t be identified as spam. At some point, you may be wasting far more time than it’s worth to keep using a phone without some major whitelist/blacklist system.
Also, what happens when the outbound calls you make are answered by AI, and you don’t know? If this AI is giving you replies that are word salad, how long are you willing to tolerate it?
I’ve been getting text messages now from companies that I actually do business with, but they are spam. Calls from companies that I have accounts with, and they are scams. At some point, SMS and phone calls will be more trouble than its worth.
And the thought of either having to go without it, the pain of replacing it, or the frustration of being strung along in a scam are not thoughts I want to have.
The internet will never be useless. There will always be a large number of sites that are not capitalist hellholes that only exist to steal user’s data or scam users or do other malicious things. This may be down to things like credit unions, federated social media, and non-profits that exist to make the world better, but there will always be something that is out there that keeps it from being useless.
There will always be a large number of sites that are not capitalist hellholes that only exist to steal user’s data or scam users or do other malicious things. This may be down to things like credit unions, federated social media, and non-profits that exist to make the world better, but there will always be something that is out there that keeps it from being useless.
No doubt that there will be people who still have morals and will run sites and services that don’t completely screw people.
But at some point, you won’t be able to tell which are legit, and which aren’t. AI generated websites can make any scam site look completely legitimate, fake thousands of testimonials, have bots post about it on every major website (Reddit, YouTube, etc.) without being caught, etc.
The currency of the internet is no longer about what’s valuable to users, but what’s valuable to bad actors, data thieves, and marketers.
There will be a tipping point when the bad far, far outweighs the good, and I’m curious to know when society decides that the internet isn’t worth using anymore.
We are well past the tipping point of malicious websites being indistinguishable from legit sites unless tou pay attention to extra informatjon like the address bar. I would say that most of the websites I used to enjoy have gone to shit between ads and ditching their better writers for cheaper staff or AI. Most of the internet is crap.
But there is stilll enough helpful content that the whoe thing isn’t worthless and I doubt it ever will be.
But there is stilll enough helpful content that the whoe thing isn’t worthless and I doubt it ever will be.
The people creating that content will fade away.
They are going to be competing with a tireless algorithm that can out out 1000x the content they can, with next to no “staff”.
Those AI content creators will be making money for someone, and legitimate content creators won’t be able to keep up unless: they use AI content creation; or have a business model that will probably result in all legitimate websites being paywalled, filled with ads, or becoming a marketing platform for brands (worse than what modern day YouTube is like).
You can just click on the three dots and on “Do not recommend” or Don’t recommend from this channel, then it will stop.
YouTube desperately needs to fix the recommendations for music. It’s either recommending me the same music I already listen to and is irrelevant to what I’m currently playing or something irrelevant to my taste and the video being watched. Think watching an Ice Cube video but having recommendations for Megadeth and Alice In Chains and Adele on the side. And if it’s not something I’ve already watched/listened to, it’s still something totally unrelated.
For the longest time, they kept recommending “Beck - Loser” to me on anything I would watch, regardless of genre. I’ve never listened to or searched for any of Beck’s music and I don’t know how that would be recommended to someone currently watching an Ice Cube music video…
And the live performances…I don’t ever watch or listen to live performances. Yet YouTube always recommends it to me. Even having a whole section dedicated to “Live Music”. Never asked for this, never searched for it…I don’t care for Live performances. I want studio only.
And what’s worse is that YouTube has a feature to tell it “Don’t recommend this video - because I watched it or don’t like it” or “don’t recommend this channel” EXCEPT for music! Why???
YouTube desperately needs to fix the recommendations for music.
I mean, I guess if someone has a YouTube account, there’s nothing wrong with using YouTube as a music recommendations system, but it isn’t really the first thing I’d think of. I mean, music isn’t really what it was designed for.
And YouTube doesn’t know what a user would listen to offline, so unless all their music-listening is from YouTube tracks…I’m not sure how representative the listening data would be of what a user would listen to.
I don’t use them, because I don’t really want to hand them a profile of me, but if I wanted to get music recommendations, I’d probably use something like Audioscrobbler, which was designed for building a profile on someone’s music-listening habits and then handing them recommendations based on that.
Do note that this is in my “All” videos recommended feed, not just filtered for music
YouTube has a recommendation system already in place and it generally works for me except for music.
I don’t use it as my main source of music, but it’s convenient at some times like putting on music on my Apple TV while I work or if I am on my computer and don’t have my offline music library available. Or if I want to see the music video because I like the video that goes with the music. I used to use it to find new artists based on my existing taste, but whenever they broke the search/suggestions is when that all stopped for me and now I just use it for the same music or if I found a new artist and want to quickly check out their music.
There’s no reason for it to be this shitty. It seems intentional on their part. Maybe making money from paid spotlights under the guise of “recommendations”.
I’m just rubbin’ and tuggin’ my fuckin’ nips
Handful of amusing and novel examples aside, generated art is an affront to han creativity and ought to be banned everywhere, or at the very least strictly relegated to its own spaces so it can be easily avoided
Hasn’t been happening to me. Then again, the music I listen to probably isn’t a big target for AI, yet.
I’m getting a lot of these! But, to me, they recommend musics from 1950s and earlier decades.
