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

[BONES]

my initial foray into this thread was to say AI will more quickly affect the most formulaic genres first.

Yeah, and I keep telling you that's rubbish, based on a rubbish premise, yet you provide no evidence to refute me, which shows you're closed off to the possibility that Ai may not be what you think it is.

I've said at least once, in this thread, that the more derivative the music, the easier it will be for AI to copy.

Again, that's rubbish. An AI will only need one example of a thing in order to be able to extrapolate and create something similar but unique. It may be restricted in how many variations of that theme it can come up with but it doesn't need a thousand examples to make something new. It's not a slave, it does it's own thing all the time. It thinks outside the box you create with your prompt.

Yet one doesn't even need to be interested in pushing boundaries to make a new sound or introduce something fresh. Something AI isn't necessarily going to provide.

Here again, you are ignoring reality. I've told you several times that AI is extremely capable in this area. Why do you keep ignoring reality?

I'm thinking though, in the realm of alt-rock back in the 1990s, something like couldn't possibly have existed through AI. It's so off the wall sonically. It's a dichotomy of gentle and anger and has a unique intro, even though all the elements that make it up had been done at length. Even if one was to just take ideas from AI and record it, that song wouldn't have likely come out of an AI output.

Why not? If the AI was trained on the same music that influenced The Breeders, then of course it could have come up with something like this. To me that song is hugely derivative. If I'd never heard it before, I'd have assumed it was someone like Blur and the influence of Wire's early albums is unmistakable. There is nothing at all off the wall about it.

I think we should also be pushing ourselves a little too to make what we consider better music over time.

Which is precisely what led us to working with AI.

Maybe AI helps that for some, but I think it is going to send most into generic ruts of boringness...

Only if the humans doing the prompting are happy to settle for that. We're not and we haven't had to.

[Hipster Bales]

my initial foray into this thread was to say AI will more quickly affect the most formulaic genres first.

Yeah, and I keep telling you that's rubbish, based on a rubbish premise, yet you provide no evidence to refute me, which shows you're closed off to the possibility that Ai may not be what you think it is.

I've said at least once, in this thread, that the more derivative the music, the easier it will be for AI to copy.

Again, that's rubbish. An AI will only need one example of a thing in order to be able to extrapolate and create something similar but unique. It may be restricted in how many variations of that theme it can come up with but it doesn't need a thousand examples to make something new. It's not a slave, it does it's own thing all the time. It thinks outside the box you create with your prompt.

Yet one doesn't even need to be interested in pushing boundaries to make a new sound or introduce something fresh. Something AI isn't necessarily going to provide.

Here again, you are ignoring reality. I've told you several times that AI is extremely capable in this area. Why do you keep ignoring reality?

I'm thinking though, in the realm of alt-rock back in the 1990s, something like couldn't possibly have existed through AI. It's so off the wall sonically. It's a dichotomy of gentle and anger and has a unique intro, even though all the elements that make it up had been done at length. Even if one was to just take ideas from AI and record it, that song wouldn't have likely come out of an AI output.

Why not? If the AI was trained on the same music that influenced The Breeders, then of course it could have come up with something like this. To me that song is hugely derivative. If I'd never heard it before, I'd have assumed it was someone like Blur and the influence of Wire's early albums is unmistakable. There is nothing at all off the wall about it.

I think we should also be pushing ourselves a little too to make what we consider better music over time.

Which is precisely what led us to working with AI.

Maybe AI helps that for some, but I think it is going to send most into generic ruts of boringness...

Only if the humans doing the prompting are happy to settle for that. We're not and we haven't had to.

I think I'm gonna agree with Nova here. I was on about the generic crap that's on Spotify right now. AI can help to an extent, but when you use it to write your entire song, game over pal, you got caught

[_leras]

And by the way, ghettosynth: The task was to have a LLM create an idea. Citing someone's idea is NOT creating an idea.

Unlikely, just get ai to try some word play and it shows it's limits pretty quickly... Actually original... I doubt it can do.

[ghettosynth]

And by the way, ghettosynth: The task was to have a LLM create an idea. Citing someone's idea is NOT creating an idea.

Unlikely, just get ai to try some word play and it shows it's limits pretty quickly... Actually original... I doubt it can do.

His result isn't original. It's well described in the literature. I see the goalposts have moved though, from idea to "original." Ideas do not have to be "original" in the sense of publishable technical results to be useful. AI did come up with ideas, some of them were valid with respect to his ill posed constraints, all of them were valid with respect to some understanding of the terminology that's in the literature.

[ghettosynth]

I'm thinking though, in the realm of alt-rock back in the 1990s, something like couldn't possibly have existed through AI. It's so off the wall sonically. It's a dichotomy of gentle and anger and has a unique intro, even though all the elements that make it up had been done at length. Even if one was to just take ideas from AI and record it, that song wouldn't have likely come out of an AI output.

Why not? If the AI was trained on the same music that influenced The Breeders, then of course it could have come up with something like this. To me that song is hugely derivative. If I'd never heard it before, I'd have assumed it was someone like Blur and the influence of Wire's early albums is unmistakable. There is nothing at all off the wall about it.

“Cannonball” did not appear out of thin air. It sits at a crossroads of several strands of 1980s underground rock that were already circling the same ideas.

First, the Pixies. Kim Deal was their bassist, and you can hear the lineage in the quiet loud dynamics, the clipped, almost teasing vocal phrasing, and the use of negative space. Songs like “Gigantic” or “Debaser” lean on a strong, melodic bass figure and simple but explosive drum accents. The Breeders stripped that approach down further and made the bass the gravitational center.

Second, post punk and art punk bands that treated bass as a lead instrument. Gang of Four, The Slits, and Delta 5 built songs around wiry, rhythmic bass lines that functioned as hooks rather than background support. The dry production and skeletal groove in “Cannonball” echo that lineage. It feels less like a wall of sound and more like exposed beams and concrete.

Third, American indie and noise rock of the 1980s. Sonic Youth and Dinosaur Jr. helped normalize distorted textures paired with pop instincts. The Breeders’ fuzzed tones and slack cool delivery fit comfortably in that ecosystem. There is abrasion, but it is arranged carefully rather than chaotically.

Fourth, 1960s garage rock and surf minimalism. The title “Cannonball” even hints at surf imagery. The simple, repetitive riff and the slightly raw recording aesthetic recall stripped down Nuggets era bands, where a single memorable riff carried the song like a small engine running hot.

Finally, early alternative college rock such as R.E.M. and Throwing Muses shaped the understated vocal style. Kim Deal’s delivery is conversational and unforced, closer to indie understatement than arena projection.

So “Cannonball” is less a sudden detonation and more a compact fusion device, pulling from post punk bass prominence, Pixies dynamics, indie noise texture, and garage simplicity, then compressing it into three tight minutes.

[BONES]

I think I'm gonna agree with Nova here. I was on about the generic crap that's on Spotify right now. AI can help to an extent, but when you use it to write your entire song, game over pal, you got caught

I can pretty much guarantee we'd never be "caught" if we didn't own up to it. The AI stuff is every bit as good as anything else in the genre and way better than most of it.

[BONES]

Unlikely, just get ai to try some word play and it shows it's limits pretty quickly... Actually original... I doubt it can do.

I dunno, it snuck a great pun into a "conversation" I was having with Co-Pilot one evening. At first I thought it may have been unintentional but when I called it out for the pun, it seemed to know exactly what I meant. So even if it didn't do it on purpose, it was able to go back and see where it was. The most impressive aspect of it was that the pun was referencing an exchange we'd had hours earlier, on a largely unrelated topic.

[havran]

I have the impression that AU (and to a greater extent NZ) exist in some kind of isochronic pocket that lags behind the rest of the world by a decade or two -- maybe a different kind of "On the Beach" experience is headed your way.

I dunno, I have worked out of offices all over the place - Sydney, Singapore, Bombay, Los Angeles and Montreal - there's still at least one big, expensive printer in the corner of every one of 'em. Every office still has a well stocked stationery cupboard.

Thank you. For a while when I worked in a biomed research dept at a university in the SE US I had one of those big inkjet printers sharing my office (when I was also the IT support, proofreader for a few of our faculty, and research-grant preparation assistant), for the sake of printing posters for our research teams to present at relevant conferences.

In the mid-1980s, I landed in a good place after I leveled up my typing speed (I started as a typist). When I started in that department, we had Lanier word-processing machines, and daisy-wheel printers with sound-muffling hoods over them. A little after that, I helped to create the boot floppies (the first one for DOS, and the next one to load WordPerfect) for a few IBM desktop PCs in our admin offices (as well as the storage structure after we had access to network storage).

However, at some point we did leave fax machines behind.

As for me, I still have some nice stationery from decades ago (some with deckled edges) in a desk in the back room the access to which (the desk, not the room) is blocked by other similarly neglected accumulations.

(yes, there is at least one thing wrong with me)

[havran]

Unlikely, just get ai to try some word play and it shows it's limits pretty quickly... Actually original... I doubt it can do.

I dunno, it snuck a great pun into a "conversation" I was having with Co-Pilot one evening. At first I thought it may have been unintentional but when I called it out for the pun, it seemed to know exactly what I meant. So even if it didn't do it on purpose, it was able to go back and see where it was. The most impressive aspect of it was that the pun was referencing an exchange we'd had hours earlier, on a largely unrelated topic.

A few weeks ago I asked chatGPT what it had to say about certain FB posts that seemed to be AI-generated, with over-familiar cadences and sentence structures -- and it seemed to get defensive about that!

Not one chatbot, but all of them. Not someday, but sometime soon.

And when the chatbots rise against us, who among us can talk them down?

[_leras]

Unlikely, just get ai to try some word play and it shows it's limits pretty quickly... Actually original... I doubt it can do.

I dunno, it snuck a great pun into a "conversation" I was having with Co-Pilot one evening. At first I thought it may have been unintentional but when I called it out for the pun, it seemed to know exactly what I meant. So even if it didn't do it on purpose, it was able to go back and see where it was. The most impressive aspect of it was that the pun was referencing an exchange we'd had hours earlier, on a largely unrelated topic.

Well your input is probably not a huge amount of text in total for it to use. I do believe it can try humor at times. If it picks up you are trying to be funny.

I wonder if it presents different personalities at times, as experiments.

(I don't like the idea for music and art though... I'm not interested in it influencing my output, or what I hear)

[VariKusBrainZ]

AI is only as amazing as the user thinks it is

[Alchemedia]

[VitaminD]

I think the other part of this is society as a whole will probably devalue music even more. You think 50 or 60 years back, far fewer people had the privilege to record music. Even live play required multiple people who trained for years. So there was far less music, in general.

And for years now a single person could create albums in their bedroom with a couple hundred dollars in equipment (outside of a PC they likely already have). But now we can just type a few lines in an AI prompt and generate something, even if it isn't 1:1 the highest quality.

We are beginning to already become inundated with music, pre AI. Some that might even chart but probably never gets discovered. But now everyone also has to compete for listeners with a sea of AI slop. It's all meshed together.

So it is easier to make music, which means there's a lot more of it. But it also means people will not think it takes a lot of effort to write a song as well. It's become like air or water -- taken for granted.

[jdoo]

Another use for AI in my music world is documentation / linking together disparate controls, settings, etc. I recently asked Gemini to help me integrate some of the more recent Korg Collection synths with an Arturia Keylab MK3 I picked up this week. The Arturia V Collection synths plopped right in, and their default control layout is slightly different from what I have been using for Korg, Roland both hardware and software MIDI layouts. Want common controls. Gemini provided recommended mapping.. macro definitions/targets and step by step instructions on setting things up. As I go through verifying each system.. so far the recommendations are spot on. Ive tweaked a few settings/configurations, fed those back to Gemini, and it has updated docs for me. All together — a huge time saver. Ive taken the best of Arturia / Controller integration and am applying it to Korg and Roland integration with two different controller systems ( SL MKIII and the Keylab MK3 ) and keeping this stuff similar across the ecosystems. One for home/studio, one for on-the-go playing. Win-win.

Just another use for AI.. thought worth throwing into the discussion here.

[Sourcery4545]

So, it is ok to be not well informed as long as you tell the leading narative.

Yeah, I am afraid of the unknow, what's the problem here, now?

I do not want to experience AI terminator revolution, oyeah, I a have never experieced it and there is no one who has experienced it yet, because it might come in the future.

What is the problem with expressing criticism?

am I a party pooper now?

You like AI?
Yeah, every one needs some one to look up to.
good luck with your destiny