If AI replaces musicians, does the entire plugin industry die with them?

[Touch The Universe]

So AI music will lack the human elements of ideas in the moment, tendencies, actuality, the process of inspiration of sounds, and the literal manifestation of the avenue of MY thoughts.

Good point, if there is a delay between input/prompt and output. What if you could change elements in fly as its playing back so there is no time delay between prompt and output. As it's playing, sing a melody to embellish what you hear etc, update, change element while in play. If you make it interactive enough and it will accept any imaginable form of engagment, voice, hand gestures, facial expressions (mood senstive). You'd be producing/creating yet without software/daw's. Very interesting... But at this point is it ai?

Why not turn the power of ai so human expression or gestures or output itself becomes the daw. Develop an ecosystem that manages human expression input itself so its no longer limited to mouse click, key pressing, etc. If it's reading your face, you could literally bypass thought itself and just allow it to read your emotions. If you hate the song, a scowl might be on your face or an eyebrow raise so it'll update and change gears or colors as one move a kaledoscope. You could use your own emotions to train it and program it.

Feed it a database of your favorite songs, songs you hate, your favorite parts of of your favorite songs. That by itself might be enough to never need any input from you ever. It could make new stuff in line with your taste, or find existing stuff you already out there. I was thinking of making an app like this actually, throw it a random 1000 songs and train it so it'll select the songs you yourself would select.

Of course, all this is realistically just an augmentation to what we have, just interesting ways or tools to add to our existing flows - if they are actually fulfilling. Imagine if you could press play and ai would generate a song that could last your entire lifetime and would be one you'd never press the "skip" button. How would that change your role/relationship to music or how you value it?

[BONES]

The point about taste I see is correct. We already have chord generators, samplers, and drum machines. I am not saying these are the same as AI, but my taste for what I choose to create with the before mentioned tools is how I apply taste. The same can be said with AI, but AI has less limitations (AI is different as a tool than all previous tools, because it removes the path of time, removing the moment element; conversely, if I use a well known hi-hat, I skip the time it would take to create one, ultimately narrowing my field of choice, but the manifestation of sound is more segmented in the moment rather than fluid in the case of AI).

I don't see how this is of any import whatsoever, nor do I think it is true. We are more than 4 months into using AI to make our next album and we're still several months away from completing it. Yes, you can generate something very quickly but generating something worthwhile still takes time and effort and provides plenty of moments, if that's important to you. To be fair. we have had all the songs we need for 6-7 weeks now but we still have a shit-tonne of work ahead of us to get them into a releasable state.

The limitation of not using AI are the choices that catalyze how my music sounds because the process is what determines a sound. For example- if I ask AI to make me a pop beat with a prompt containing X, Y, and W, it will take all data and previously made sounds, as well as use its imagination to create a result that will be Z. If I do everything manually using older tools, the process will take me down a different route based on my individual taste in choices, so the result will be different.

That's only relevant if you accept the first thing the AI spits out. If you have specific goals, be it style or sounds or rhythms or whatever, you can spend a lot of time with Ai getting those things right, trying out all kinds of variations to see what's going to work best for you. To me it has felt little different from doing it any other way, you still have to work at it to get to a result that you're happy with.

If I know every single choice I want to make before exploring the music process itself down to the exact Hi Hat and AI can translate it, then the result will be whatever I input in that moment

No it won't because the AI won't get it right the first time. In fact, if you know exactly what you're after, it probably won't ever get there because it will always put its own spin on whatever you ask from it. It works best if you only have a vague idea of your goal, as you normally would when you're writing a song from scratch or from a single idea. Instead of looking for elements to fill out the idea into a song, you give AI instructions as to what kinds of elements you want and let it have its head. You get the best from it when your prompts are detailed but not specific, if you know what I mean.

removing any imaginative choices during the process

Not at all. The AI's imagination is different from yours but that can be a good thing, in that it takes you down paths you might never find on your own.

so I want to get that straight since it is similar to using loops in this manner).

It doesn't feel like that to me. A loop is a loop, it will always be the same, whereas one of the things I like about the AI output is that it changes through the song - no 4 bar sequence is ever exactly the same. The AI tends to inject a lot more variation that I ever would, which I appreciate.

Now, if I keep running prompts over time until it creates something more of my "own", now the result will be more to what I would create if I didn't have AI, but it would still not match entirely what I would have created. So AI music will lack the human elements of ideas in the moment, tendencies, actuality, the process of inspiration of sounds, and the literal manifestation of the avenue of MY thoughts.

Not even close. You still all of those things, it just does it in a different way. You still make all the final choices, it's more like having a crazy assistant who never quite does exactly what you tell him or her to do but if you keep at them, it will eventually give you something like what you intended. You can get your "inspiration of sounds" by going through 400 Hive presets and 50 Battery kits looking for things that work or you can let the AI find those things for you. It's one of the best aspects of using AI for me - it comes up with great sounds almost every time. As for YOUR thoughts, I think you badly over-rate their value.

Can AI do it's own thing in regards to this? Sure, but the output will always be a translation, never a living breathing output of my personal being. So is what I said right or semi-accurate or am I completely off?

Mostly off. You might find the fact that it's not slaved to your personal quirks and limitations liberating. It might open up a whole new avenue for you to explore and make your own, as it has done for us.

If you hate the song, a scowl might be on your face or an eyebrow raise so it'll update and change gears or colors as one move a kaledoscope. You could use your own emotions to train it and program it.

Or maybe it will take that as a cue to make the song angrier. Your face is never going to be a reliable tool.

Feed it a database of your favorite songs, songs you hate, your favorite parts of of your favorite songs.

You can do that now but it doesn't necessarily take much notice of it. And how is that any different to how AI is trained anyway?

That by itself might be enough to never need any input from you ever. It could make new stuff in line with your taste

If that's all you want, it can do that now. We've got hundreds of songs we're happy to listen to but only a handful we feel are good enough to put on a album with our logo on it.

[Touch The Universe]

Indeed, even as is, it requires finesse, tweaking to get things suitable to taste -- and creativity is still creativity and fulfilling as long as that is there - at least theoretically. If that improves in the future with real time adjustments, there conceivable won't be a need for a daw. Your mind and prompt and taste itself become the DAW. It then just becomes another way to get from a to z.

Record live stuff with many people old . Use DAW to make band with one guy tweaking parameters or keys with HANDS, modern way. One guy tweaking with prompts NO HANDS needed but the playlist, ideas, happen in ones head, potential future way. The progression seems to require less "mechanical" effort from many bodies to one, until the mechanical side itself can be replaced with though. Creativity might become therefore metaphysical.

Learning daws might only be necessary to familiarize oneself with the ways in which a sound can be tweaked synthesize, filters, types, effects. How will you know what to even prompt unless you know the layout of the land so to speak. But again, this too is shallow thinking - just show it an example of what you are after and you don't need to know the type, etc of distorted. Ear training, taste, mind itself becomes the skills and "software" and daw itself.

That said, no that I'm typing this, there is something pleasurable about knowing these little tricks and ways to tweak sound. You could concievable say in real time to an ai, open cutoff filter bp high rez 80% slowly, or fast, or relative to input keytracking etc or whatever randomized way. I think hand gestures would be the missing key. I won't need a mouse if I could use expression to control the cutoff somehow, in a very detailed way. I digress. That's why these thiings are interesting to think about, ai poses many questions on creativity itself and what's pleasing about it. The things that cause pleasure can shift an morph, but at its core its still expression and detailed complixity to othewise be wrapped up in an engaging and immersive experience itself. Preserve these elements and one might begin to see synths layouts and daw themselves limiting to this ultimate end - a fulfilling flow state that is immersive and engaging that can be preserved and shared with others.

[Touch The Universe]

Having said that, I'll play devils advocate to the opposing side now. We are both mind and body, physical and metaphysical. To go to other extreme one might be lacking. There is ideally an optimal balance between hands on stuff and metaphysical prompting. Ideally, again, I'd love to see ANY CONCIEVABLE tedious, mundane, trivial thing, that is time consuming to be mitigate metaphysically with thought and prompts. The pleasurable, hands on, even going back to analog groups of folks making music together, even with ai prompts, sophisticated, expressive tweaking to be preserve the optimal pleasure and joy here as well. That is the ultimate goal, no, maximum joy, pleasure, and happiness with the least amount of barriers to creativity and flexibility.

Heck, why not just have a dance mode. Tweak the song in real time by the dynamics of the group dancing. Those who know more about the ways to manipulate the song via expression could affect more tweakings. If competing requests, weighted priority on uniqueness, or unity. Haha, lots of funny ideas are possible in these line of thoughts.

[enroe]

Heck, why not just have a dance mode. Tweak the song in real time by the dynamics of the group dancing. Those who know more about the ways to manipulate the song via expression could affect more tweakings. If competing requests, weighted priority on uniqueness, or unity. Haha, lots of funny ideas are possible in these line of thoughts.

Haha, yeah yeah - that's guaranteed to happen: People will dance to an
ancient country dance group - but the music is completely AI-based and
even changes while they're all dancing live.

But the question is: Do we want that? Do we really want to allow that?

[Touch The Universe]

Heck, why not just have a dance mode. Tweak the song in real time by the dynamics of the group dancing. Those who know more about the ways to manipulate the song via expression could affect more tweakings. If competing requests, weighted priority on uniqueness, or unity. Haha, lots of funny ideas are possible in these line of thoughts.

Haha, yeah yeah - that's guaranteed to happen: People will dance to an
ancient country dance group - but the music is completely AI-based and
even changes while they're all dancing live.

But the question is: Do we want that? Do we really want to allow that?

Indeed, if it's boring, do some gestures to change it up, too happy, change to a minor key for depth etc.

As a "Mode" why not? Do we always want to be initiating all times, no. Sometimes, just sit back and go with the flow. Do we always want to prompt, no, just a tool. Do we always want to tweak, hopefully not, lol.

[Zeisner]

But! You are improvising, just playing on your piano without knowing what will happen and all of a sudden you repeat playing a cool melody or chord progression. Not sure, whether AI can already do it.

It can't because it has no ears, no senses, no brain, no body. It does not even have a short- or long-term memory like humans. It's nothing more than a lossy compression codec.

[vurt]

So AI music will lack the human elements of ideas in the moment, tendencies, actuality, the process of inspiration of sounds, and the literal manifestation of the avenue of MY thoughts.

Good point, if there is a delay between input/prompt and output. What if you could change elements in fly as its playing back so there is no time delay between prompt and output. As it's playing, sing a melody to embellish what you hear etc, update, change element while in play. If you make it interactive enough and it will accept any imaginable form of engagment, voice, hand gestures, facial expressions (mood senstive). You'd be producing/creating yet without software/daw's. Very interesting... But at this point is it ai?

Why not turn the power of ai so human expression or gestures or output itself becomes the daw. Develop an ecosystem that manages human expression input itself so its no longer limited to mouse click, key pressing, etc. If it's reading your face, you could literally bypass thought itself and just allow it to read your emotions. If you hate the song, a scowl might be on your face or an eyebrow raise so it'll update and change gears or colors as one move a kaledoscope. You could use your own emotions to train it and program it.

Feed it a database of your favorite songs, songs you hate, your favorite parts of of your favorite songs. That by itself might be enough to never need any input from you ever. It could make new stuff in line with your taste, or find existing stuff you already out there. I was thinking of making an app like this actually, throw it a random 1000 songs and train it so it'll select the songs you yourself would select.

Of course, all this is realistically just an augmentation to what we have, just interesting ways or tools to add to our existing flows - if they are actually fulfilling. Imagine if you could press play and ai would generate a song that could last your entire lifetime and would be one you'd never press the "skip" button. How would that change your role/relationship to music or how you value it?

have you tried "playing an instrument"?

[Tubeman]

Record live stuff with many people old . Use DAW to make band with one guy tweaking parameters or keys with HANDS, modern way. One guy tweaking with prompts NO HANDS needed but the playlist, ideas, happen in ones head, potential future way. The progression seems to require less "mechanical" effort from many bodies to one, until the mechanical side itself can be replaced with though. Creativity might become therefore metaphysical.

No one gonna believe you made anything.

[BONES]

Does that matter? Is it important to you that everyone knows you made it? Isn't it enough that what you made is worthwhile in and of itself? Isn't that where the satisfaction comes from?

It can't because it has no ears, no senses, no brain, no body.

But it understand those things much better than you or I ever will.

It does not even have a short- or long-term memory like humans. It's nothing more than a lossy compression codec.

It does. In fact, it keeps meticulous records of our interactions with it and recalls with perfect clarity. It also retains access to all the training data it consumes. After all, what is RAM but short-term memory and what is a hard-drive but long term memory?

[Aloysius]

AI can be incredibly useful at analyzing and understanding things. For example, I asked ChatGPT why I thought a certain songwriter wrote what come across as “cinematic lyrics.” It was able to break down the text and explain how those effects were achieved. Of course, it still needed my input to generate the output, but it was impressive how it could help me understand the process.

Has anybody else used AI to understand or analyze something that seemed mysterious at first? I’d love to hear your experiences!

[Touch The Universe]

1000 word prompts - go deeper - scary useful Thats why im so frreaked and immidiatley look to the bigger picture of things - whats the end of these things - like icarus flying to the heavens - get burned

[BONES]

None of our prompts are anywhere close to 1,000 words. Mostly they'd be less than 100, I reckon. It won't ever do precisely what you ask it to do so there is no point trying to be too specific with anything.

[Touch The Universe]

The quality of the output depends on how you prompt. Sometimes you have to use negative prompting—explicitly telling the AI what not to do—and requesting a specific format. I've even written prompts over 2,000 words long to get it right.
Remember, prompts aren't just requests; they can be knowledge shares. You can provide everything you know about a topic and then ask the AI for fresh insights, to identify gaps, or to point out weak spots. Simply sharing information provides context that helps the AI give better feedback. Tell it what you know first—you’ll be surprised by how much more useful the response becomes.

[BONES]

The thing is, the model that the AI uses will only have been trained to recognise a specific list of words. It may or may not have any idea what you're asking for. Anyway, it's not designed to be precise so you'd be wasting your time with a 2,000 word prompt. You need to think of your prompts as suggestions, not instructions. i.e. "I'd like something a bit like X, possibly with some of Y, maybe in a Z style", not "I want X plus Y plus Z".