At least currently, Jordan Rudess's use of AI is fairly acceptable. Like you said he's just having fun with it, which is totally fine. He had a discussion with Devin Townsend where they both agreed that AI is a tool, not a replacement, which is the best use scenario for AI. Similar to your sampling example, if someone used AI for music inspiration, heard something they liked and expanded on it, I think that's fine, it's really no different compared to hearing a great riff on a song and doing your own unique spin on it. The problem is when people don't do their own take, they just, well.... take, and then try to sell it when they didn't actually *do* anything.
200k views and 9k likes is actually pretty good bud. That is a 22:1 ratio and is definitely in the range for popular content, where many very popular videos get 25:1 or worse.
Also, talking about Dream Theater, there is some, I don’t want to say evidence, but some definitely some tells that Hugh Syme might’ve used some AI for the artwork. The Night Terror single art has some goofy stuff going on with it - the door knob is on the wrong side, and the girl in the art has really weird hands. The main album has a few things going on, but nothing that’s too bizarre. Ugh, I love Dream Theater too, and I hope it isn’t the case.
The theory that I and a lot of other people have is that the album cover is not one whole AI image, but rather contains AI elements that were photoshopped in. AI has a very distinct look to it, it looks inhumanly perfect (because it is) and the Parasomnia cover definitely has visible imperfections if you look at it for a few seconds. And I'm not talking about the lighting and the hands and stuff. The elements aren't as cleanly separated as they are in most AI imagery, which leads me to believe Hugh Syme is at it again with his photoshop nonsense. Individual parts of the cover (the girl, the window, etc) definitely look like AI, but they look more like they were generated separately and then edited in. Hugh notoriously has a history of using stock images in his artwork, and with the massive abundance of AI stock images right now, it's very likely that he just found some AI stuff online while looking for material and threw it all together in an ugly photoshop mess the way he usually does nowadays. I don't even know if he did this intentionally or not. The guy is 70 and it's possible that he didn't even recognize it as AI. Or maybe he knew exactly what he was doing and just didn't care that it was AI. Maybe they weren't stock images at all, and he generated the individual elements himself. Who knows?
I mean, on a purely technological level, it's pretty impressive what AI can do, and one of your commenters below tried really hard to make a good case for certain types of AI music, but I still maintain there's no heart and soul in it and there are still massive ethical concerns about the source material (there's a vast gulf of difference between "I learned to play an instrument so I could write something reminiscent of a favourite artist" and "this neural net, despite being well-trained, still has to pull sounds from sources not belonging to the person training it"), and honestly, it IS going to put actual musicians out of work. Except for one thing. I don't care HOW well-trained your neural net is or how ethical your AI is. It ain't touring. No one's buying concert tickets to Udio or Soma or ChatGPT :P And another thing, I suspect a LOT of music, especially pop music and anything "mainstream", is already making use of AI in some form, whether it's ChatGPT or Soma writing lyrics or some other AI working on beats/chord progressions/etc. In fact, I practically guarantee at least some of your favourite pop music of this year has had AI involved in its creation in some format or another. There's no way you can tell me that an industry that's as image and profit-focused as the pop music industry isn't looking for ways to cut costs, and that they wouldn't at least consider the use of AI tools to streamline the process. There's a very good chance one of those half-dozen songwriters on every pop song is someone who just fed a prompt into an algorithm... not that I want to be "POP MUSIC BAD" or detract from a genre you love, but I think there are some objective truths about the economic reality of the music industry that makes something like this very plausible. Maybe this is just "Tinfoil Hat Travis" speaking, though. I dunno, it seems pretty believable to me.
The recent Beatles single, though, is one of the few uses of AI in music that I *DO* approve of, where they managed to isolate and clean up John Lennon's vocal from a demo tape recorded decades ago... there was little additive about what AI did to his vocal, it just simply separated the instrument sounds out and cleaned up the sound quality of what was left, while actual recordings of George from a 1990s attempt to finish the song were added to new tracks recorded by Paul and Ringo. AI didn't create anything for it, just used as a tool to clean up some recordings... and I think that would be one of the few ethical AI uses in art.
I've stumbled upon an AI "RPI inspired" album called "Il Giardino dell'Impossibilità" at Bandcamp. Some people had already purchased it. For me it's another level compared to YTB "fictional band / album" videos, and the creator (Gerd Weyhing) has his own music portfolio - several CD's listed at Discogs. So I'm wondering, what's the difference? Where is the line? He argues working with AI is creative process, which requires a lot of decisions, and the final product "is full of emotion". I'm just not sure...
My last comment on a video of yours was really long, so let's write another book. Again, apologizes in advance: I want to note that I am very opposed to generative AI, but I know how a myriad of these generative AI tools work because it's right now, unfortunately taking over a large part of my job at the moment. One of the things that people do get wrong about generative AI is that everything that it produces is aesthetically "bad" - and I am not playing devil's advocate here, it's certainly not all 'bad' (big italics here). It's uninspired, it's just content, it's often bizarre, it's you know quote, 'slop', and a million other things. But I've heard plenty of real musicians make worse things than some (big italics here) of the things AI has made. I think that's perhaps the bigger worry, and not in the way that most people think: If generative AI isn't plateauing, that is to say, if it's going to continue getting better by orders of magnitude, I think that will actually disrupt the music (and art) industry in a more profound way than it already is. This relates to the comment you make at 13:45: Because while we as humans take inspiration from artists, we can't copy their 'brush strokes' (or voice) so to speak with computer precision. It's the places where you're limited in addition to the places that you're exceptional as an artist that make you an artist. The external experiences that aren't related directly to your art that influence it. The human artist experience isn't just input > output. It's the space in between. The issue is that there are already, right now, a few AI voice cloning tools that allow you to combine multiple voices you've trained together into a single one. UDIO and SUNO can already mimic the artists you mentioned fairly well. This is not 20 years in the future problem, this is a now issue. Because you can download those programs right now for free. I wouldn’t be surprised if we see the exact thing you’re describing happen in the next year or two, because we’re there. I would say to smaller, up and coming artists too: You might not be able to pay for the exact cover art you want. You might not be able to have things mixed the way you want. But changing the way we make art to value speed over your own artistic growth and experience will be detrimental to you in the long run. You will not be able to navigate your own artistic impulses if you depend on them being automation. I fear we're about to enter a real artistic recession soon if we can't convince people the merits of being creators, instead of just slop curators and consumers.
I dunno man, I've never heard a good song made by AI. At least not one I'd care to listen to start to finish. The fact that a lot of humans make a lot worse doesn't make it not bad. You can call it technically competent, but that's not the same as good.
Didn't Steve actually do the instruments for the song tho? If there's an actual human element involved in the process then I don't see a problem. Lots of ppl are using Soma/Sona(?) to create songs for lyrics they wrote. I don't see a problem with that, and they probably could copyright those songs as well.
I wrote a comment (long and detailed) about AI music. Why was it deleted? It included no links, no swear words, etc. It took me a lot of time, very frustrating.. :(
@@NotesReviews main ideas were: AI music has many forms and is not always "Create me a song...", it may include thousands of creative decisions. Check how Udio works. You can control every second of the generated output. Also it is not just "combining pieces of information from different sources", the training of neural nets more similar to a training of real humans in a way that you get an "inner represenation" of what is music, and then use it (in a combination with "creative decisions" to create output). In this way every person who listened to Beatles at least once is influenced by them and should be banned from creating music the same way, that "the neural net, trained on Beatles songs".
You can check "Red Vector - Летопись Страны Советов", progressive rock concept album about the history of Soviet Union, created by a non-musician with Udio (including thousands of "creative decisions" and several weeks in creation time), which would not be possible without AI. Many (thousands) of people listened to it and many liked it. Author expressed his feelings and thoughts about the mentioned theme with music, and no real musicians were injured in process. Isn't it a good thing?
Now the quality of AI tracks is still not perfect, but even now people (including musicians) sometimes could not distingush between "Udio-generated" and "real" music. AI music is still in its infancy, Suno came out less than year ago, and Udio - even later. What will be in a year from now? Nobody knows. But it will not be just a toy for "funny meme songs" or "get me a new Rammstein song about Easter"-like things, especially with projects like Udio, which is intended to deepen the possibilities of a "controlled music generation" - with every new version they add controls, which could influence the output. AI is a tool, which could be used not only by a non-musicians to express themselves (or just have fun), but also to enhance the ability of real musicians to generate music, which they could not before because of different limitations (money, ability to get specific instument of vocalist with a specific style, time limitations, etc.)
At least currently, Jordan Rudess's use of AI is fairly acceptable. Like you said he's just having fun with it, which is totally fine. He had a discussion with Devin Townsend where they both agreed that AI is a tool, not a replacement, which is the best use scenario for AI. Similar to your sampling example, if someone used AI for music inspiration, heard something they liked and expanded on it, I think that's fine, it's really no different compared to hearing a great riff on a song and doing your own unique spin on it. The problem is when people don't do their own take, they just, well.... take, and then try to sell it when they didn't actually *do* anything.
Notes, are you into finnish prog like Haikara, Wigwam, Tasavallan Presidentti?
200k views and 9k likes is actually pretty good bud.
That is a 22:1 ratio and is definitely in the range for popular content, where many very popular videos get 25:1 or worse.
Also, talking about Dream Theater, there is some, I don’t want to say evidence, but some definitely some tells that Hugh Syme might’ve used some AI for the artwork. The Night Terror single art has some goofy stuff going on with it - the door knob is on the wrong side, and the girl in the art has really weird hands. The main album has a few things going on, but nothing that’s too bizarre. Ugh, I love Dream Theater too, and I hope it isn’t the case.
The theory that I and a lot of other people have is that the album cover is not one whole AI image, but rather contains AI elements that were photoshopped in. AI has a very distinct look to it, it looks inhumanly perfect (because it is) and the Parasomnia cover definitely has visible imperfections if you look at it for a few seconds. And I'm not talking about the lighting and the hands and stuff. The elements aren't as cleanly separated as they are in most AI imagery, which leads me to believe Hugh Syme is at it again with his photoshop nonsense. Individual parts of the cover (the girl, the window, etc) definitely look like AI, but they look more like they were generated separately and then edited in. Hugh notoriously has a history of using stock images in his artwork, and with the massive abundance of AI stock images right now, it's very likely that he just found some AI stuff online while looking for material and threw it all together in an ugly photoshop mess the way he usually does nowadays. I don't even know if he did this intentionally or not. The guy is 70 and it's possible that he didn't even recognize it as AI. Or maybe he knew exactly what he was doing and just didn't care that it was AI. Maybe they weren't stock images at all, and he generated the individual elements himself. Who knows?
I mean, on a purely technological level, it's pretty impressive what AI can do, and one of your commenters below tried really hard to make a good case for certain types of AI music, but I still maintain there's no heart and soul in it and there are still massive ethical concerns about the source material (there's a vast gulf of difference between "I learned to play an instrument so I could write something reminiscent of a favourite artist" and "this neural net, despite being well-trained, still has to pull sounds from sources not belonging to the person training it"), and honestly, it IS going to put actual musicians out of work.
Except for one thing. I don't care HOW well-trained your neural net is or how ethical your AI is. It ain't touring. No one's buying concert tickets to Udio or Soma or ChatGPT :P
And another thing, I suspect a LOT of music, especially pop music and anything "mainstream", is already making use of AI in some form, whether it's ChatGPT or Soma writing lyrics or some other AI working on beats/chord progressions/etc. In fact, I practically guarantee at least some of your favourite pop music of this year has had AI involved in its creation in some format or another. There's no way you can tell me that an industry that's as image and profit-focused as the pop music industry isn't looking for ways to cut costs, and that they wouldn't at least consider the use of AI tools to streamline the process. There's a very good chance one of those half-dozen songwriters on every pop song is someone who just fed a prompt into an algorithm... not that I want to be "POP MUSIC BAD" or detract from a genre you love, but I think there are some objective truths about the economic reality of the music industry that makes something like this very plausible. Maybe this is just "Tinfoil Hat Travis" speaking, though. I dunno, it seems pretty believable to me.
The recent Beatles single, though, is one of the few uses of AI in music that I *DO* approve of, where they managed to isolate and clean up John Lennon's vocal from a demo tape recorded decades ago... there was little additive about what AI did to his vocal, it just simply separated the instrument sounds out and cleaned up the sound quality of what was left, while actual recordings of George from a 1990s attempt to finish the song were added to new tracks recorded by Paul and Ringo. AI didn't create anything for it, just used as a tool to clean up some recordings... and I think that would be one of the few ethical AI uses in art.
I've stumbled upon an AI "RPI inspired" album called "Il Giardino dell'Impossibilità" at Bandcamp. Some people had already purchased it. For me it's another level compared to YTB "fictional band / album" videos, and the creator (Gerd Weyhing) has his own music portfolio - several CD's listed at Discogs. So I'm wondering, what's the difference? Where is the line? He argues working with AI is creative process, which requires a lot of decisions, and the final product "is full of emotion". I'm just not sure...
My last comment on a video of yours was really long, so let's write another book. Again, apologizes in advance: I want to note that I am very opposed to generative AI, but I know how a myriad of these generative AI tools work because it's right now, unfortunately taking over a large part of my job at the moment.
One of the things that people do get wrong about generative AI is that everything that it produces is aesthetically "bad" - and I am not playing devil's advocate here, it's certainly not all 'bad' (big italics here). It's uninspired, it's just content, it's often bizarre, it's you know quote, 'slop', and a million other things. But I've heard plenty of real musicians make worse things than some (big italics here) of the things AI has made.
I think that's perhaps the bigger worry, and not in the way that most people think: If generative AI isn't plateauing, that is to say, if it's going to continue getting better by orders of magnitude, I think that will actually disrupt the music (and art) industry in a more profound way than it already is.
This relates to the comment you make at 13:45: Because while we as humans take inspiration from artists, we can't copy their 'brush strokes' (or voice) so to speak with computer precision. It's the places where you're limited in addition to the places that you're exceptional as an artist that make you an artist. The external experiences that aren't related directly to your art that influence it. The human artist experience isn't just input > output. It's the space in between.
The issue is that there are already, right now, a few AI voice cloning tools that allow you to combine multiple voices you've trained together into a single one. UDIO and SUNO can already mimic the artists you mentioned fairly well. This is not 20 years in the future problem, this is a now issue. Because you can download those programs right now for free. I wouldn’t be surprised if we see the exact thing you’re describing happen in the next year or two, because we’re there.
I would say to smaller, up and coming artists too: You might not be able to pay for the exact cover art you want. You might not be able to have things mixed the way you want. But changing the way we make art to value speed over your own artistic growth and experience will be detrimental to you in the long run. You will not be able to navigate your own artistic impulses if you depend on them being automation. I fear we're about to enter a real artistic recession soon if we can't convince people the merits of being creators, instead of just slop curators and consumers.
I dunno man, I've never heard a good song made by AI. At least not one I'd care to listen to start to finish. The fact that a lot of humans make a lot worse doesn't make it not bad. You can call it technically competent, but that's not the same as good.
Didn't Steve actually do the instruments for the song tho? If there's an actual human element involved in the process then I don't see a problem. Lots of ppl are using Soma/Sona(?) to create songs for lyrics they wrote. I don't see a problem with that, and they probably could copyright those songs as well.
I wrote a comment (long and detailed) about AI music. Why was it deleted? It included no links, no swear words, etc. It took me a lot of time, very frustrating.. :(
@@EvgenySmirnovVersus :( it never came though
@@NotesReviews main ideas were: AI music has many forms and is not always "Create me a song...", it may include thousands of creative decisions. Check how Udio works. You can control every second of the generated output. Also it is not just "combining pieces of information from different sources", the training of neural nets more similar to a training of real humans in a way that you get an "inner represenation" of what is music, and then use it (in a combination with "creative decisions" to create output). In this way every person who listened to Beatles at least once is influenced by them and should be banned from creating music the same way, that "the neural net, trained on Beatles songs".
You can check "Red Vector - Летопись Страны Советов", progressive rock concept album about the history of Soviet Union, created by a non-musician with Udio (including thousands of "creative decisions" and several weeks in creation time), which would not be possible without AI. Many (thousands) of people listened to it and many liked it. Author expressed his feelings and thoughts about the mentioned theme with music, and no real musicians were injured in process. Isn't it a good thing?
I did too :( youtube gave me an error and it still says 9 comments despite there only being seven. Curse you youtube
Now the quality of AI tracks is still not perfect, but even now people (including musicians) sometimes could not distingush between "Udio-generated" and "real" music. AI music is still in its infancy, Suno came out less than year ago, and Udio - even later. What will be in a year from now? Nobody knows. But it will not be just a toy for "funny meme songs" or "get me a new Rammstein song about Easter"-like things, especially with projects like Udio, which is intended to deepen the possibilities of a "controlled music generation" - with every new version they add controls, which could influence the output. AI is a tool, which could be used not only by a non-musicians to express themselves (or just have fun), but also to enhance the ability of real musicians to generate music, which they could not before because of different limitations (money, ability to get specific instument of vocalist with a specific style, time limitations, etc.)
Excellent points all succinctly discussed ❤️❤️❤️