Some thoughts on AI music, fandom, and making things for yourself

https://suno.com/s/psi97E2ZvRQjiWRx

https://suno.com/s/Jny3bVQla36DoZSa

I made these two songs, and I wanted to make a post about how I’m feeling more comfortable with AI-generated music.

A lot of people have talked about whether AI music is most useful and proper for making funny songs on family road trips about small events that happen in the moment. I would expand that to songs nobody’s ever actually going to write, sing, or put any production value into.

For me, once a week I meet up with a D&D group. We don’t necessarily publicize it because people might get a little upset, but our characters are made with AI image generation because:

• It’s not like we’re going to be spending real money on art for a D&D game, other than maybe buying stories.
• We might reuse the art in the future, but not in any serious public or commercial way.

It’s a little bit throwaway and just some background, but I know people would be super upset about that. I think people are more uncomfortable with AI music than almost anything else. I can definitely see that when these things get added to Spotify or released outside of a small friends group, that hits a little closer to the problem people are having, where it can actually take money away from real artists. I 100% get that.

One of the things I’m finding most useful is creating very personal soundtracks. I’m with Mike on this. I really enjoyed doing DJ-type stuff. I loved just sitting for hours with song tracks, making samples, and then doing remixes. Back in the day, that was what I spent most of my time doing, because I was either administering Linux servers, writing PHP code, or mixing music. That was pretty much what I did online when I wasn’t maybe jumping into an online Quake mod game like Team Fortress, the original.

I wasn’t a good map maker, so I couldn’t really contribute to the game. I wasn’t a good enough coder to contribute to any of the mods.

I’m probably a better singer and guitar player today than I ever was back then, but I still wasn’t good enough to do more than just play and sing along with my own playlists. To create things, I would do mixing, and I still have a desire to do that. I would love a production setup where I could take samples and also mix in stems of instruments, where I kind of sing out the beat a little bit and create drum lines that I could add to my own mixes.

For me, most of this isn’t for release to the public or Spotify, or to make money. I’ll be honest, most people probably wouldn’t even like my mixes. Not a lot of people are probably into crossovers of Beastie Boys and Lenny Kravitz, though I will say it sounds pretty freaking cool when you do that: a little “American Woman Sabotage,” if you know what I mean.

Just like D&D games where people sit around and make up their own stories, I’m sitting around writing my own lyrics and having AI expand on them or help me with rhymes. I’m really getting into non-rhyming music. I’ll be honest, I’m kind of going into a little more modern style of poetry and mixing in some arcana from the books I read. I’m finding that I really enjoy making things like specific character songs.

I’m a huge Dune fan, and nobody is going to make a Chani or Scytale soundtrack, but I did. I’ve read the books, I support the artists, and I think it’s really neat to make something that can fit the character.

While I won’t say I’m spending days on every song, a few of them I do spend days on. For some of them, I’m really spending an hour or two going back and forth on lyrical changes, getting fresh ideas, and going over where I think it sounds correct. I’m also going over different musical styles and trying to get different instrument mixes, which is kind of the same thing I did when I sat down to mix music.

I can sit down and listen to the books, which I’ve certainly done. I’ve listened to the entire Dune series and grabbed new books when they come out. I’ve gone through the first 20-some books at least five times.

What’s fun for me is to sit down, remember a character, remember one of their iconic scenes, start describing it, start working out some different lyrical points, and then find the correct mood for them.

That’s kind of a way for me to be a little participatory in my fandom, but again, it’s not being released to the public in any major way. I think things like fan-made films are interesting. There have been some really good book-based Star Wars fan films with high production value. Those are really cool, but I’m not working with anybody. I can’t act, and I have no intention of actually trying to make any money or anything. Nobody’s ever going to find these songs unless they’re specifically looking up a Dune character and they’re just super fans of the book in the same way I am.

I’m one of the people who read the Frank Herbert biography, and there’s just not a whole lot of people who are going to go watch Dune 3 and think, “You know what? To make sure I fully appreciate the film, I went and read the Frank Herbert biography.”

I think one of the things I wanted to do, with all the negative talk out there, is highlight some of the good personal uses of even some of the most publicly disliked AI out there. I don’t feel like I’m in bad company when you’ve got people like the Spielbergs of the world and Ron Howard talking about how AI could have a positive impact on their creative process. I’m going to be honest: I feel a lot less apologetic about making AI music, especially when it’s for myself.

• I’m not trying to copy any specific artist.
• I’m not trying to put things on Spotify.
• I’m not trying to use anything like licensed samples.

I’m just sitting and listening on my own. Now, I suppose the other argument could be made that because I’m listening to AI music, I’m not listening to real artists, which is true. If I am in the mood for the music that I made, I am not in the mood at that moment for other artists. But it’s not as though I would stop doing that and start listening to other music. I will almost always listen to the music I am actually interested in.

So I brush aside the argument that this is taking away from them, at least in the personal-use case I’m talking about. I’ll also make another point: a lot of people are trying to have their cake and eat it too. People are still going down this path of “it was trained on everyone’s data,” as though cover artists aren’t doing something at least spiritually similar. I am a huge fan of Maphra. Maphra has thus far done covers using other people’s lyrics and other people’s backing tracks, though it’s in a karaoke style, not just pulling from them and putting her own spin on it. I think that’s 100% cool, but even that makes people mad. Those same people are such purists that, obviously, it’s not just AI they’re against. It’s cover artists too.

They’re probably the people who are still angry at Vanilla Ice. Vanilla Ice was never going to destroy any of the income or profit-making of David Bowie and Queen. Simply for the fact that if you were a fan of the Vanilla Ice song, you had most likely not been old enough to hear the original David Bowie and Queen song. Technically, he introduced new listeners to David Bowie and Queen. I would argue that the crossover went in one direction. It was not Vanilla Ice pulling in David Bowie fans, lol. It was kids listening to Vanilla Ice and going, “Oh wow, this is a cool beat,” and then going and listening to the original song.

I also think about the story around “Bitter Sweet Symphony” by The Verve. That song is iconic from my teenage and young adult years. They were sued by The Rolling Stones’ side over the sample, and the lead singer, with one of the most viral songs ever, didn’t make a dime on that song for years after the settlement. The lawsuit pretty much took the profit from him and some from the label as well. The only money he made from it for a long time was from live performances.

I would argue that some works are transformative, and whoever decided to sue so hard as to actually destroy another artist’s vision is the old-school example of the kind of thinking people now apply to AI. They would be happy to destroy any level of creativity in favor of what they view as someone with more creativity, as though The Rolling Stones never heard The Beatles and thought, “That little guitar lick sounds good. Let’s play with that and make it more heavy rock.”

I just wanted to point out again that, for me, I’m making this as a serious creative effort that has to be correct. The mood has to be right. I’d probably go through a dozen iterations on a lot of these before I get the right sounds, and that’s not even including fixing the lyrics.

I’m going to push back hard because I think the double standard here is that the same people who are anti-AI music are perfectly fine with code, as though programmers have no creativity and don’t put learning, expertise, and craft into their work. I’m not saying every programmer is a craftsman, but I really don’t like the fact that people are crapping on one end of AI and not another.

I disagree with them on most of it, probably 90% of the argument. To pick on music and visual art and say that musicians and painters and people who draw are so much more special than programmers is just not right. Not to be rude, but I’ve got some words for that, and they’re not kind words, lol.

Those same people are not upset if you use your phone to scan an object and then 3D print it. There’s no AI involved. You’ve got to replace this weird shroud on a car part, so you print off something that’s heat-resistant and put it onto your car.

Few of those people are actually thinking about the industrial designer who made that part and the company that put real work into designing and implementing that part. I’m not saying the person using their phone to scan or 3D print is bad. I love both of those things. But people are very selective about who they crap on and what, to them, counts as pure human creativity. I would argue that those people are not very creative in their thinking. They’re very one-track and simple-minded.

To say that someone is not creative when they’re creating an exhaust system for a rebuilt 1980s Mustang GT, but someone singing off-key on an acoustic guitar in their bedroom is somehow the apex of creativity, is just ludicrously wrong. Both people can be using their creativity to the maximum, and both people are likely doing those things not in an economic sense. They’re not suddenly going to start an exhaust company, and the person on the guitar is not going to become the next YouTube-famous Ed Sheeran. They’re focusing on an area and trying to do cool things in that area, and they can’t do that without a human mind, regardless of what tools are helping them.

It goes back to my handcrafted argument. Everyone talks about handcrafted music, but most of them don’t think about it. Most of the instruments they start with were made on an assembly line, for the most part. I played trumpet, and I’m pretty sure the metal for my trumpet was not hand-smelted in a forge in an epic display of classical metallurgy.

I’m not saying I’m defending AI in every use. I am defending people using AI in a non-slop way. Someone might listen to my songs and say, “That’s a bunch of slop,” and that’s fine. To them, it can be slop. To me, I listen to it and I don’t care one whit if they like it or not.

That’s also kind of one of my own ethical lines that I don’t have to struggle with. If I’m not really doing a big public release, I don’t feel like I need to keep these songs secret. If someone finds it and wants to add it to their role-playing catalog of interesting fantasy music to listen to, that’s cool. The fact is, I don’t care if other people like it. If every other person in the world hated the music I’m making, I wouldn’t care one little bit.

If I’m not past that line, I feel zero guilt about using these tools. Were these products trained on music? Yeah. So was every other musician.

As wonderful as Mozart might be, give him a string and a piece of wood, and if he had never heard music, he would not have been remotely as good as he was. Without his contemporaries, without the contemporary instruments, without the buildings and the architecture to play in, we’re all standing on giants.

And just a newsflash for everyone who’s all about that: nobody’s compensating Mozart. Whatever Mozart learned from, he didn’t compensate those people in the way we now talk about compensation. In fact, sheet music at the time was handmade and highly expensive, and copying was rampant and common. But we’re not living in those bad old days, thank goodness.

It’s like architecture. Imagine someone saying they’re going to be a carpenter but they’ve never seen a building. That gets at the idea that I’ve been trained on everything else, and I get that it’s a different type of training than human training. It’s also like the people who value hard work for the sake of hard work.

I slightly agree with that. I think hard work and even manual labor are character-building. I’m glad my son followed in my footsteps and has done real physical work, not just relying on others to do the work for him. But some people think that makes the work better. If that were true, we would all have ugly furniture. Somebody would have worked really hard and poured their heart into furniture without the skill, but because they worked so hard, we would value that furniture more. It’s sort of the IKEA effect.

What we actually value is someone who isn’t just working hard, but has gained expertise and is working really well. We value their skill more than their hard work. Nobody really cares if Mozart worked hard. They care about his skill. I think it’s interesting when we hear about deaf composers of the past still composing music long after their hearing is gone, but I wouldn’t call that the same thing as working harder. They just had so much skill that nothing could stop them, and that is cool and inspiring.

I think, overall, people seem to be more in favor of the “handmade human” when it seems artsy, but suddenly, if it has any practical use, they see it as utilitarian. For a simple, basic argument, music and paintings don’t have a practical use, and I would argue that that is a completely incorrect take. But let’s say, for the sake of argument, they see programming as purely utilitarian. That means they’ve never seen a game that a single person made and paid attention to every pixel and every reaction. Those people can’t see the beauty in something like the original Mario Brothers or Prince of Persia, or even something more modern and cartoony.

I find it hilarious that people are upset about the use of AI in game asset creation, but not one drop of ink seems to be spilled about AI programming for the coders.

Not that I would just call them hypocrites, but I think their take on the code might be right, and they may wish to extend that to asset creation. In other words, I don’t think they’re hypocrites for hating it all. I think they’re hypocrites for not accepting it all as a tool.

I would agree with them that a one-shot prompt to create a game isn’t going to be super interesting or worth writing home about. I think we value the novel, and it still requires a human to get novelty out of things. The character songs I make would be nothing without the works of Frank Herbert, Brian Herbert, and Kevin J. Anderson.

On the other hand, there’s also the economics of it. Someone could be mad that I’m making and listening to AI music, but I genuinely enjoy it because it’s the only alternative. Nobody’s making music for me. There is no actual “me” channel, regardless of how many AI DJs they make. Those songs were made for mass consumption, for the most part, unless you really get into B-side tracks.

The other economics is that I could spend a year finding and testing instruments, singing, and editing to finally make a single song. That will not make me value it more. It will simply mean that, rather than having 150 songs on different things in a year, I will have a single song, and it will actually not be nearly as good.

Will I have learned something? Sure, but I was already a musician before. I play the trumpet, the piano, and the guitar, and I sing. I don’t do any of them particularly well enough to be in a local band. Going through the process, would I become a better musician? No. I spent my entire youth doing audio mixing and playing those same instruments. I am as good of a musician as I will ever be, full stop.

I think a lot of people in their early 20s think that you just continually get better over life, and we like to say that because people can still pursue their passions into age. I’m going to be really honest here: my musical talent is never ever going to improve in any major way. The only thing I’m trying to do is get a little bit better at a couple of singing techniques that are not going to make any difference to any other listener at all. Adding a little more grit in my high range is barely going to be noticeable by anyone but me and will probably sound worse than just straight singing. Yeah, there’s no getting better for me. It’s just not happening.

In fact, this post I’m making now and going to share on Discord is going to be made with AI. It’s not right now. Right now, I’m probably 30 or 40 paragraphs in, drafting my thoughts with a recorder into a text box that I’m then going to tell AI to clean up, organize a little bit more, and make look interesting for a blog post that I’m going to share on Discord.

Right now, the only AI working is the transcription of my voice to words, speech-to-text, so it’s 100% me right now. It’s going to become something a little better: something where I don’t feel I am trading my life’s time for something not meant to make money, but only meant to get a point across and help communicate to other people.

I’m literally 25 minutes into speaking and thinking this out. God help me if AI is going to add hyphens or repeat a certain adjective or adverb, because apparently that destroys all prior work. Or something like that. I think that’s the point of this.

I think there are two different questions:

• The use of AI in industries
• The use of AI by people, doing people things to make those people happy

I think one of the first things we could do is keep the argument focused on the companies and such, and not go after the people. No offense, but there are like four people on Earth who even have room to criticize the creative choices of Spielberg or Howard for their directing, and that’s a list that includes people like Francis Ford Coppola. I think they’re trying to change the industry zeitgeist in the same way that people can still try to improve themselves and do these things. They’re not wholly abandoning all effort without AI. I would almost say it’s an augmentation issue.

If you can’t see, an AI could give you not just sight, but extra sight.

There’s a neat TED talk by a woman who had lost her legs, and she talked about the advantage of having artistic artificial legs. Being able to choose her height for the day. Some days she feels like being 6’5”, other days she feels like being 5’7”.

I don’t mean transhumanism. I mean augmenting and extending people.

Everyone deplores the use of autotune, but if you’re just a bad singer and you enjoy singing, you can sing into something and it sounds good to you. When it comes out for other people, you throw some autotune on it, and it sounds to them like it sounds to you. Not gonna lie, making fun of that person or being angry at that person is kind of like being angry at Stephen Hawking for using a vocalizer.

Am I considering myself disabled? No, absolutely not. That’s ridiculous. But I would say I’m a less-abled singer and a less-abled musician. I’m not as good as other people. I’m better than some and way worse than others. I enjoy making things that sound good to me, even if I don’t necessarily have to listen only to myself. I do that plenty.

It’s not as though I stop doing things that aren’t going through AI, but if I want to make something repeatable and I want it to look good and be an actual representation of what I had in my head, then yeah, I’m going to go to AI the same way I might go to a dictionary when my creative spelling is incorrect or my grammar might be a little off.

My point of view is that I’ve gone down a road a lot of people haven’t. I’ve gone down the road of publishing and academic research, literally spending a year poring over every period, every citation, every turn of phrase, removing specific words, going over hundreds and hundreds of pages of notes, and producing a document hundreds of pages long to be put out publicly for the specific purpose of being criticized by the entire scientific establishment. I’ve already done that work, and there’s a type of work that is kind of handcrafted because it has to be so exceptional that it has to be above basic reproach.

But I don’t feel like writing a dissertation every time I just want to post some thoughts online or make an interesting song idea. I could sit here and tell you what would sound cool and kind of describe a picture in a story, and maybe you can imagine the song in your head, but you’re not going to hear what I had intended. That’s where AI comes in. Once I can hear it externally, when I talk to my son who writes his own music or I talk to my wife or I talk to you guys, I can say, “Here it is. Here’s what I had in my head.” AI also added things that I didn’t think of that are actually even a little bit better, and I decided to keep them in.

I’m not going to make the AI taste argument. I think talking about human taste is a little simplistic. I think it’s far more complicated than that. But it is editorial choice. If AI is wrong, then the editors of The Atlantic and The New York Times are using their own writers as AI if we’re going to go by that logic. I think there is definitely something to be said about curation and editorial choice that separates what most people would consider AI slop from what other people actually value when they make it with AI.

Anyway, those are just my thoughts.