Paying lawyers is only for big companies like Sony.
I remember Shakira plagiarism

AI is more complex, is like a training right, not a copy right.

 

while music does get plagiarized often, this example is a bit of a stretch imo. the melody in question is a very common vocal technique. this could have easily been done with zero plagiarism. The songs are completely different, its just the oh oh oh oh melody, and yah, im sure theres countless other songs with the exact same melodic phrase.

 

And so on.
You got to the point.
With AI is even more difficult than this case.
The AI is not even copying.
The AI is trained.
There is a law of Copyright not of training right.
Sony is just the next Kodak.
Kodak case was even more sustainable, because they bought patents of digital photo.

 

It's not AI, it's machine learning, which means it's incapable of producing anything on its own by learning how to produce. This is a very important distinction.
And no, it is a copyright issue - they use a produced piece as a "stamp" on a noise file, then applies slight variations and then you get the result. It's basically a top down approach which is really easy to prove in regard of music(music and music shapes\forms don't exist in nature, it's purely abstract form of art, unlike most of visual arts).

So, it's not complex and what they're doing is essentially another trick, nothing esle. Like I've mentioned above, it produces nothing, just slightly varies an actual piece to certain degree and that's it.

This is one of the main reasons you have so many layers of legal bs, basically paid "partisan" marketing(like pieces in Rolling Stone and so on), so many complex layers of registration of invsetments and ownerships and so on. The PoSes that do this know exactly what they're doing and that it's just another froad.

But what pisses me off the most is the fact that they're wasting resources on pointless activity, because machine learning, applied correctly, could help actual music producers dramatically. There are so many possibilities to decrease presence of "dumb work", repeatative actions, organizing process of varications and so on... Instead we are getting this dumb and useless shit.

 

lmao no
It's not even close to the Kodak case. What these guys are doing is literally stealing, it's not a figure of speech or exaduration in this case. This is what they do - steal. Imagine a "deisgner's t-shirt production pipeline" - there's a cotton production(companies that grow cotton, collect it, process it and so on) and then they sell it to fabric producers; fabric producers... produce fabric and sell it to t-shirt producers; t-shirt producers make "base" models and sell it to t-shirt designers and whoever needs base models; designers put their logos, arts, accessories, colors and so on; then those guys sell it the stores that sell designed t-shirts from countless number of t-shirt designers.

What the PoSses that do this "AI" music do is they just take t-shirts from designers and sell them as if they represent the entire pipeline of production when they didn't even design it, claming it's all their work and of the "magic machine" that does everything and even more(!), the machine that they designed because of science and stuff(basically "AI").

It's gonna die really soon. I'm almost certain about it.

Btw a small fact for doomers: read how much energy those "learning machines" consume and remember that an actual AI(who's possibility of creation in the first place is still under question lmao!) will consume much, much, much more. For example, Google servers consumed 22.29 terawatt hours(!) in 2022 alone, and it's without the rest of the stuff they consume energy for, all of this is simply for storage and basic CPU calculations. I think we can deduce the rest from this fact on our own.

 

Are you kidding?

Things and time are never ever going backwards.

Technology is always a path only forward

I remember when people were against online music distribution, that by the way it was far more controllable than AI. It was a disaster for the music industry, but it was inevitable

People don’t understand that, even when we don’t like it, it will happen.


AI is here, the genie is out of the lamp.

 

Last edited: Jun 25, 2024 at 7:13 PM

Year 2012... Kodak sues Apple, HTC over digital image patents
https://www.reuters.com/article/idUSN1E8090AA/

Kodak loses in suit against Apple, RIM
https://www.nytimes.com/2012/07/23/business/kodak-loses-patent-lawsuit-against-apple-and-rim.html
https://www.reuters.com/article/idUSN1E8090AA/

 

The problem here is that there is no effective way to stop this. Let's consider the best-case scenario: AI-generated music is banned from being trained in the United States. The effect of this would be that AIs would be trained outside the United States, so there would need to be some kind of control over the origin of the music that shows its creator and how it was made. This type of control is extremely rigid and difficult, an strict regulation on the origin of content that would only be valid for the United States. The rest of the world would continue with the technology, and AI would probably develop elsewhere, in another country.

 

My friend, stady basics of the subject you're arguing about first. AI doesn't exist.

 

It was just a matter of time. I found it annoying when I heard some middle-aged person saying that the quality of things were deteriorating... until I became a this middle-aged guy! . But I think it goes much further than that. It has to do with good habits, healthy habits, or better said, the lack of them...

Things are getting easier and more instantaneous to be produced and packaged, by anyone, but for those who want to make a living from music, for example, no, because you should earn better for what you do. But I don't see that in the near future, unless you come up with something really innovative and cool, but today, sincerely, simply because it seems like general listeners are all drugged and anesthetized. I'm starting to think that that exposure to short forms of media and the trivialization and equating of things (like everyone doing BRAAAM, or the ball of the moment, or even trying to sound like A and B exclusively: where are C...Z? or A+Z, H+Y+U, for example?) just fucked with the millennial generation's head.

 

Last edited: Jun 26, 2024 at 1:37 PM

I think there is a lot of tidying up to do here...

Machine learning is a subfield of AI. So to say it's not AI because it's machine learning is nonsense.

IIt does nothing on its own, as it is dependent on an instruction. So far your statement is correct. However, these AIs produce. In the case of Udio, Suno, etc. - waveforms. And " producing" is meant here entirely in the sense of the meaning of the word. Just because the AI's work steps are different from yours doesn't mean that it's not production. The result is the same, a waveform. No more and no less.

Whether it is a copyright issue or not is better left to the courts to decide.

That's a very simplified way of describing it, yes. And an abstract one, too. But the problem with simplifications is that they are always inaccurate. And the abstraction to the technical level is also not particularly helpful for basic understanding.
The problem with this statement is that it gives the impression that a song is used as input, then slightly modified and then output. Some even have the idea that the models contain samples that are simply mixed and reassembled like a puzzle. But neither is the case.

I would rather use an anthropomorphising example to explain the situation like this:
Let's assume you are an AI. I show you a lot of pictures of people. Men, women, children, in all kinds of poses. There are descriptive terms underneath each picture: Man, sitting, thinking. Muscles, arm, man. Breast, woman, etc. All the pictures can be viewed publicly and were taken by someone. I didn't ask the person who's in them or the photographer whether they agreed to me showing you these pictures. I didn't give you the pictures either, I just showed them to you. You look closely at all these pictures and read the descriptions. You have no idea what you are seeing or what these terms mean, but over time you begin to recognise patterns and mentally categorise what you see. You have no concept of human, man, woman or certain activities such as walking, standing or sitting. For you, these are all just forms in various forms.
One day I come to you and say: "Here is a large, rectangular block of granite with smooth surfaces. Do the following: Man, woman, dancing, happy. You have one hour." You put your hand on the granite and it starts to change shape.
When I come back an hour later, there is no longer a rectangular block of granite. You have changed the shape of all the sides, but the result doesn't look anything like a dancing couple. "Another hour," I say. I return after an hour and inspect the work. "Another hour." It goes on like this a few more times. After each additional hour, the granite block looks less like a granite block and more like a happy couple dancing. After 30 hours, I stop the process. In front of me is a sculpture of a man throwing a happy woman into the air on his hands. The woman has a few too many fingers and one of the man's legs is sticking out at a strange angle. "Impressive," I say. All these details. That's what we exhibit."
A little boy is at the exhibition and laughs: "What kind of idiot did this? That ugly woman has far too many fingers."
A guy called Jon Feltheimer walks around the sculpture and examines it from different angles. "Theft!" he suddenly cries out. "This is theft! This pose is clearly from Dirty Dancing! We didn't give anyone permission to use images from our movie!" I try to calm the man down: "Well, first of all, this is not Dirty Dancing. Secondly, it is not forbidden to look at screenshots of Dirty Dancing. Thirdly, the two people in the sculpture resemble neither Baby nor Johnny. In fact, they don't resemble anyone in our image database. And fourthly, we don't even know if the AI made up the figure from learning from a screenshot of Dirty Dancing, or from another image with a similar figure, or came up with it on its own. The AI does not copy images. It looks at images and learns patterns and shapes."

You, on the other hand, are standing next to it and don't understand a thing. You have looked at a lot of pictures, learned a lot, your artificial neurons have made a lot of connections and these neurons react differently to different inputs. You did what you were told. You used these connections to create something new. You don't have any pictures to copy from. You only have artificial neurons that perform a function.

What may be pointless for you at the moment may not be pointless for others. And you said "applied correctly", what do you mean by that? Can you give an example or two?

Okay... Who was something stolen from? When? And what?
Here's a little denouement task:
Think of the Mona Lisa. Try to visualize her.
Are you picturing her right now? Try to imagine even more details. How is it now?
Do you have Mona Lisa clearly in your mind's eye?
Is that theft?

After all, the Mona Lisa should be in the Louvre and not in your head. Nobody allowed you to make a copy. And yet Mona Lisa is in your memories right now.

What if you put a picture of Mona Lisa on canvas (assuming you can paint)? Is that theft then?

After all, the painting is still hanging in the Louvre. Chances are you've never been to the Louvre. And that you have never made a copy. You have seen a photo of the painting somewhere, your neurons have made connections and because you have seen Mona Lisa very often, you have a very clear image of the Mona Lisa in your mind's eye.

The medium doesn't play a role here. Be it pictures, videos, music. It's the same principle.
And it's not much different with artificial neurons. Don't get me wrong. It's not the same as with a living being, but it's similar to the principle.

Naah.... I don't think so. The potential is too great. It promises technical progress in pretty much all fields, plus power interests and economic interests. And to be precise, it hasn't even really started yet. For a lot of power-hungry people, it's a wet dream come true.

Well... I don't know why you're directing this argument specifically at doomers, because the danger potentially posed by AI has little to do with power consumption, but that's not how something like this works anyway.
ENIAC consumed about 150 kWh per hour (this is about the average power consumption of 1 person per month). About 20 years later, a simple pocket calculator with more computing power than ENIAC needed no more than a photocell to operate.

 

Last edited: Jun 27, 2024 at 8:41 AM

Whether you personally agree with the term "artificial intelligence" or not is irrelevant to the fact of the subject's existence.
I suspect that you are not satisfied with the assignment of the term "intelligence" to artificial intelligence and deny the existence of intelligence to the subject. Well... Here's the thing:
1. there are a whole range of very different definitions of intelligence and it has never been possible to agree on a single definition in intelligence research. In order to discuss whether something is intelligent or not, it would therefore first be necessary to clarify which definition each discussant is referring to.
2. there are several sub-fields in intelligence research, i.e. several types of intelligence (depending on model). And while machines have already completely outperformed humans in some types of intelligence, they are far inferior or not existent in others.

 

Last edited: Jun 27, 2024 at 8:12 AM