Rick Rubin on AI (& now Graeme Revell, too)

[CinningBao]

Why the f**k would you say that? What sort of arrogance makes you think you can judge the value of any piece of music you hear?

If you don't like a genre, or if you are unfamiliar with it, how can you tell what a good song in that genre is? What would you compare it to? If I played you a song, would you know if it sounded more like Leaether Strip or Skinny Puppy? Numb or Dance or Die?

I never said anything like that; you've extrapolated nonsense from my paraphrasing of your insults.

I've heard a shit-ton of Country music in my time but I don't have the first f**king clue what makes a good Country song. It all sounds flat and boring to me. So if someone put up an AI generated Country song and told me it was amazing, I'd have to take their word for it. With our stuff, there might be a handful of people here who'd get it but the rest of you wouldn't know if it was good or not. You certainly wouldn't have the depth and breadth of knowledge on the subject to know how original it is, or how derivative, which is central to this discussion.

The difference in our appreciation of this technology seems to be that, regardless of the emotional/monetary judgement of the material, I think most artists who are looking to express their genuine joy or pain, recognise their soul is most exposed when they work with there hands on an instrument and, sometimes, sing at the same time. Or just play something and write on top of it. The principle of farming creativity to a machine makes me wretch quite frankly. I'd rather listen to some old blind guy from the 1920's playing a 4 stringed guitar and expressing the pain of his existence rather than an AI's 'idea' of that pain. Being able to tell the difference isn't a concern to an artist; it's a concern of the listeners and most artists don't write for the audience; you write FOR YOURSELF.

Do you, for example, understand why our current album is no. 2 on the German Alternative Chart this week? If you went away and listened to it half-a-dozen times, do you think you'd understand any better? Sure, you can assess the technical aspects of it - how well mixed it is, etc., -but that's not at issue here, nor is it what makes something popular.

I lived in Germany for a while so I do have a sense of the culture and why certain things hit there and nowhere else. Having a quick listen to the material you're talking about, I'd say that it's hitting over there because they always have had a lot of industrial 'mechanistic' genres in the charts. Brash hiphop, the entire krautrock scene.. perhaps the vibe of the music matches how the language they use 'feels' to them? Perhaps the language is simple enough for them to follow? I noticed that as a feature of popular English-language songs over there; the language was very simple. They all have to learn English in their studies but the general working culture tend to forget most of it as they live and work in Germany.

It's the same with the stuff I listened to on your site. For all I know it could be chart-topping, million selling stuff within its genres but to me it sounds generic and boring, so I wouldn't be able to judge whether an AI generated song in the same genre was better or worse because neither thing connects with me.

EXACTLY. YOU'RE impression of my drawly, dry 70's AOR crap is that it sounds generic and boring becuase you don't listen to that music (I'm dangerously assuming) so the subtlties are lost on you. As they probably would be on someone else who had only listened to, for example, hard house for the last 20-30 years. Flip it around, and the subtleties of your music are entirely lost on me. It sounds like angry machines trying to make sense of themselves. I mean, I can listen to it as a producer and appreciate the sound selections, the interplay of the drums and bass.. but I don't 'get it'. It doesn't make me feel "ah, this human being also understands my pain of existence".

It's just one of those things. What makes it work, whether it connects with you or not, is completely subjective, which means you need to be into it to understand. Because what's central to this discussion is if an AI generated song can invoke the human emotion of a song written by humans. There is no insult anywhere in that.

No direct insult for sure, but it feels, to me, like people who use these regurgitating brains to 'create' aren't creatng for the fun and explorative nature of creating, they're doing it for spead, or laziness, or lack of knowledge..

And I'm not saying that to insult you, I'm saying it because its part of my music 'religion' or belief system or whatever you want to call it.

I get it. You could be clicking random chord generators instead and doing fundmentally the same thing; it's about choosing what to use for your thing. But you'd be clicking probably 100 more times to get something useful from randomness than a brain which has been trained on millions of other people's songs where you can choose from, say, 4 reasonable choices.

Maybe it's too late for you (in your mind) to learn to play and write to the level these machines are doing it for you; I will always disagree with these conclusions - we can ALWAYS learn something new. How about the joy of making mistakes? those brains will never make a wrong chord choice, but we, as humans, can easily miss-hit a chord or timing position, and unwittingly add some sense of humanity into the music. THAT'S what AI 'writing' brains won't ever do. And to me, that's what makes good music. Errors. Small, but errors nonetheless.

There's a 'forced neatness' to your sound I can't deal with. Its like it's telling me I have no choice but to lock in with the energy and just let it take me on this angry journey. I've never gelled with music which sounds like it's trying to push it's way into my mind. I like to feel it understands my condition (as a human) and it's giving me a chance for it to 'sit' with me. But we all have different experiences of life, and how we relate to the people and systems around us, whether we feel in control or out of control of our worlds.. and music cannot hide those narratives. I like music to feel like a hug, and I'm geussing you don't. That probably sums all this up.

[Michael L]

^^ Thanks for such a thoughtful post, Cinning Bao! You capture in words the many ways AI will always fail as an artistic tool. I was particularly struck by the importance of human errors. I consider errors as sometimes like hidden doorways into secret passages.

{ Producers like Rubin or animation composers like Revell are not singer-songwriters, so AI may in fact help them deliver product }

[jancivil]

I don't see it as any more or less fraudulent than clicking notes into a piano roll with a mouse.

That doesn't even quality as a thought. You're a f**king idiot.

[jancivil]

I consider errors as sometimes like hidden doorways into secret passages.

YES. You have to know when not to correct. I've had some wild ones, I don't get there from intentionality.

[VOODOO U]

Quite humorous. KVRists talking about the beauty in human errors. A place where sequencers arps and pitch correction software (Melodyne for example) is commonplace.

[VOODOO U]

Oh and just for the record, you don't have to utilize any A.I. output directly. I've been taking bits and pieces i like and playing/recording those parts myself to make a finalized song.
It's called borrowing. It's what rock stars have been doing fior the last six thousand years.

[BONES]

I'd rather listen to some old blind guy from the 1920's playing a 4 stringed guitar and expressing the pain of his existence rather than an AI's 'idea' of that pain.

Here's the thing, if that guy was able to express himself and the AI has been trained on it, then the AI will be equally capable of expressing those same emotions when you ask it to. That's how it works. It may not understand but you will, so you can direct it to express exactly what you want it to. It doesn't need to understand what it's expressing any more than that guy's guitar does. How is that not completely obvious?

Being able to tell the difference isn't a concern to an artist; it's a concern of the listeners and most artists don't write for the audience; you write FOR YOURSELF.

So? The AI can write for me better than I can. There's nothing special about anything I do, I can assure you. But that's OK because I don't live in my head, I live in the physical world and it's the physical impact of music that works for me, not the emotional pish-tosh. Music for me is catharsis, not therapy. I write for the mosh pit, not for the romance section of the public library.

I lived in Germany for a while so I do have a sense of the culture and why certain things hit there and nowhere else. Having a quick listen to the material you're talking about...

Yeah, so you'd be guessing, which is my point. It's hard to assess the value of something if it's not something you're into.

YOU'RE impression of my drawly, dry 70's AOR crap is that it sounds generic and boring becuase you don't listen to that music (I'm dangerously assuming) so the subtlties are lost on you.

The difference is that, mostly because it is so innocuous, I get exposed to that kind of music far more than you're likely to be exposed to my kind of music. As I said, it's the kind of stuff they play in the change rooms at work, in bars and restaurants and probably in shops, too. I also have a mate who is right into it and I have to go and see his 9 piece band play now and then, just to be supportive. It's all covers - earlier Stevie Wonder, James Brown and that kind of stuff. I don't enjoy it at all but I get free beer, so I go.

Flip it around, and the subtleties of your music are entirely lost on me. It sounds like angry machines trying to make sense of themselves.

Which is exactly why I don't see the point in putting any of the AI stuff up for anyone to listen to. The test will be next month, when we play a couple of them in our set and see what the audience reaction is. My prediction is that they will go off but you never know.

mean, I can listen to it as a producer and appreciate the sound selections, the interplay of the drums and bass.. but I don't 'get it'. It doesn't make me feel "ah, this human being also understands my pain of existence".

Then you definitely wouldn't connect with out music, because the reaction we want is for people to jump around and scream at the top of their lungs, Network-style - "I'm as mad as hell and I'm not gonna take it any more!". We want people to get angry, to question the pathetic nature of their pointless existence within society and ask why their lives can't be so much better.

No direct insult for sure, but it feels, to me, like people who use these regurgitating brains to 'create' aren't creatng for the fun and explorative nature of creating, they're doing it for spead, or laziness, or lack of knowledge..

First up, given the type of music you're writing, how is yours not simply a "regurgitating brain", rehashing music from 50+ years ago?

I've never thought of song writing or production as fun. I see it more as a necessary chore, like doing the laundry or washing the dishes. You seem to rely on other people feeling the same way, people happy to pay you to write music so they don't have to.

If you want to have songs to get up on stage and perform, you need to make some. I'd have stopped 30 years ago, when I finally had a full set, except I/we keep getting better at it, so new songs seem to be a good idea that keeps us relevant to an audience. But I enjoy playing covers at least as much as I enjoy playing our own songs. In a way, the AI generated songs feel a bit more like covers but I think that's more about the process than the way they connect with me.

I get it. You could be clicking random chord generators instead and doing fundmentally the same thing; it's about choosing what to use for your thing. But you'd be clicking probably 100 more times to get something useful from randomness than a brain which has been trained on millions of other people's songs where you can choose from, say, 4 reasonable choices.

Are we talking human brain here or AI? because I don't see any fundamental differences in that regard. We are all the product of our accumulated knowledge. Everything we are we have learned along the way. e.g. You have your 70s style R&B (or whatever you want to call it), which is something you've assimilated through your lifetime, so you know how to make more stuff in a similar vein. In the same way, if you ask an AI to do the something in that genre it will have all the same music, which it has been trained on, to inform what it does. Which of that stuff it chooses to work from will be down to how you guide it.

Maybe it's too late for you (in your mind) to learn to play and write

I have absolutely no interest in it. It would be a complete waste of my time. I have nothing to prove to anyone, I don't care if you or anybody else looks down on us because of our working methods, for the simple reason that the process is irrelevant. Only results matter and, right now, we are getting the best results by using AI, so that's what we're doing. However, it already seems that a newer model of Tunee has poisoned the well, so it's unlikely we'll use it again, unless they fix it. After all, it's not like we haven't written chart toppers on our own.

I will always disagree with these conclusions - we can ALWAYS learn something new.

We/ve learned how to work with AI, I think that fulfills the brief far better than rehashing music from the 70s, regurgitating techniques that have been around forever. Broaden your horizons, try something different and learn about different processes.

How about the joy of making mistakes?

Obviously I'd have no idea about that, nor do I have any desire to experience joy.

those brains will never make a wrong chord choice

Are you sure? I've heard some pretty bad stuff come out of Tunee. It is anything but infallible.

but we, as humans, can easily miss-hit a chord or timing position, and unwittingly add some sense of humanity into the music. THAT'S what AI 'writing' brains won't ever do.

Actually, that's EXACTLY what AI does, all the time. If you give it exactly the same prompt 50 times, you'll get 100 different results. It seems to be incapable of giving you the same thing twice. Most of the good stuff we're getting out of it is from just those kinds of happy accidents, where it hasn't quite done what you've asked it to or it's taken your prompt too literally and done something you weren't expecting with it. That's what make sit interesting to work with.

The song I finished last week is an amalgamation of ideas from three different versions of the same prompt. Each added something a little bit different and combining them made for a better result, so that's what we've done. They were all at different tempos, so we couldn't have combined them if I hadn't rebuilt it in the DAW without losing quality.

And to me, that's what makes good music. Errors. Small, but errors nonetheless.

Then you'll love working with AI. Seriously.

You capture in words the many ways you desperately wish AI would fail as an artistic tool.

There, I fixed that for you.

I was particularly struck by the importance of human errors.

Why are "human" errors any more significant than other kinds of errors? The opportunities for those kinds of serendipitous diversions when working with AI are far greater than when doing your own thing. The challenge is to keep them to a minimum so you don't stray too far from your original goal. Or maybe it does something awesome and you run with that instead, then come back to your original idea later.

Producers like Rubin or animation composers like Revell are not singer-songwriters

You clueless fucktard. Graeme Revell was writing songs 50+ years ago and he's scored feature films for at least a dozen Oscar nominated directors. Here's a song that got them on the telly in 1983 -

[BONES]

I don't see it as any more or less fraudulent than clicking notes into a piano roll with a mouse.

That doesn't even quality as a thought. You're a f**king idiot.

Then perhaps you can explain the difference, please? I'd be keen to hear it.

[enCiphered]

This reads like a cynic who treats emotional distance as proof of intelligence rather than a limitation. What a creative bankruptcy..
Anyway, none of this explains why the best take is always the one you forgot to record and no AI yet capable of finding the missing cable that was just here.

[CinningBao]

Here's the thing, if that guy was able to express himself and the AI has been trained on it, then the AI will be equally capable of expressing those same emotions when you ask it to. That's how it works. It may not understand but you will, so you can direct it to express exactly what you want it to. It doesn't need to understand what it's expressing any more than that guy's guitar does. How is that not completely obvious?

But were the brains you're using trained on anything but perfection? I don't think so. I tried recently to prompt a badly sung and played instrument and it couldn't do it. I wanted some human error, something which sounded like 'instinct' but it couldn't. It only understands whats it's been trained on. The AI can't replicate emotions; it doesn't UNDERSTAND emotions, all it knows how to do is regurgitate its contents to satisfy the request as best as it can, based on how it understood the input. And you're comparing that to a man with a guitar? They are so far from each other its wild that you see them as the same thing.

So? The AI can write for me better than I can. There's nothing special about anything I do, I can assure you. But that's OK because I don't live in my head, I live in the physical world and it's the physical impact of music that works for me, not the emotional pish-tosh. Music for me is catharsis, not therapy. I write for the mosh pit, not for the romance section of the public library.

If you really perceive that the AI can write better than you, then I can't argue with that. We only know what we know. It just doesn't feel very 'punk' of you to farm your feelings to a machine. Maybe that IS punk? Steampunk?

Yeah, so you'd be guessing, which is my point. It's hard to assess the value of something if it's not something you're into.

Well not really, that's what I was saying. I can see quite well why your music energy works better over there than, say, here in the UK. The simplicity of language, the hard-edged energy, it almost sounds like the language. So I can well understand why certain songs of yours hit better over there.

The difference is that, mostly because it is so innocuous, I get exposed to that kind of music far more than you're likely to be exposed to my kind of music. As I said, it's the kind of stuff they play in the change rooms at work, in bars and restaurants and probably in shops, too. I also have a mate who is right into it and I have to go and see his 9 piece band play now and then, just to be supportive. It's all covers - earlier Stevie Wonder, James Brown and that kind of stuff. I don't enjoy it at all but I get free beer, so I go.

I should really have released some of my other music. I don't only write insipid 70's love rock. Here's some badly recorded snips from a gig I did a decade or so ago

First up, given the type of music you're writing, how is yours not simply a "regurgitating brain", rehashing music from 50+ years ago?

How is your music not trying to evoke similar emotions to the punk-era of around 50 years ago but with an electronic ‘edge’?

I've never thought of song writing or production as fun. I see it more as a necessary chore, like doing the laundry or washing the dishes. You seem to rely on other people feeling the same way, people happy to pay you to write music so they don't have to.

And therein lies the pinchpoint. I actually enjoy just sitting down and noodling on a piano or guitar, now THAT's cathartic. You're looking to achieve, as you said, some kind of recognition of power within the people to be angry at (society? government? other regular shitty people? money systems? bus services? ..) the shit around them. And maybe to empower these people to make the changes which make them happy? Maybe that's the bit I feel is missing from your sound. It just remains angry; it's not looking for peace, which is what I do with my music.

If you want to have songs to get up on stage and perform, you need to make some. I'd have stopped 30 years ago, when I finally had a full set, except I/we keep getting better at it, so new songs seem to be a good idea that keeps us relevant to an audience. But I enjoy playing covers at least as much as I enjoy playing our own songs. In a way, the AI generated songs feel a bit more like covers but I think that's more about the process than the way they connect with me.

That, to me, is a more honest way to appreciate what you're doing with it.

I get it. You could be clicking random chord generators instead and doing fundmentally the same thing; it's about choosing what to use for your thing. But you'd be clicking probably 100 more times to get something useful from randomness than a brain which has been trained on millions of other people's songs where you can choose from, say, 4 reasonable choices.

I will always disagree with these conclusions - we can ALWAYS learn something new.

We/ve learned how to work with AI, I think that fulfills the brief far better than rehashing music from the 70s, regurgitating techniques that have been around forever. Broaden your horizons, try something different and learn about different processes.

Actually, that's EXACTLY what AI does, all the time. If you give it exactly the same prompt 50 times, you'll get 100 different results. It seems to be incapable of giving you the same thing twice. Most of the good stuff we're getting out of it is from just those kinds of happy accidents, where it hasn't quite done what you've asked it to or it's taken your prompt too literally and done something you weren't expecting with it. That's what make sit interesting to work with.

I'm sorry, I literally can't see any similarities between choosing which AI generated audio from a single prompt you prefer to sitting with an experienced guitar player and trying different ideas (not that I do that since I play the guitar) but I think you can see what I"m saying. And, again, I'm fairly sure you'll disagree.

The song I finished last week is an amalgamation of ideas from three different versions of the same prompt. Each added something a little bit different and combining them made for a better result, so that's what we've done. They were all at different tempos, so we couldn't have combined them if I hadn't rebuilt it in the DAW without losing quality.

Isn't that more remixing of ideas? Which could be argued IS the basis of creating. But if your process makes you happy, do it.

Then you'll love working with AI. Seriously.

Ok, I'll put my cards on the table; I've been working with and training audio brains for the past 6 years, literally hearing them try to make sense of the source material. I want to say it's clever and amazing, but it's just numbers. A thing looking at a short bit of a wavecycle it regurgutated, trying to work out the most likely place for the next sample-point. That's LITERALLY all it is. It has no awareness of itself, or the sample points it chose to give me prior to 'the next one'. It's dumb. Looking at it from the perspective I have, you really can see how good companies and govs are at telling the masses that AI thinks and ponders, so you should respect it. ALL IT WANTS TO DO IS GIVE YOU THE NEXT MOST LIKELY <sample point/word (or actually letter)/adjustment to noise (when it comes to images> thing it thinks you want to see/read/hear.

I'm looking at releasing a product which, in a way, takes the 'smartness' out of AI generation. No prompts, no meta data, just pure unadulterated vector-based queries because I think people need to hack these brains like the old russian scientists did to our soft fleshy synapse bundles in the 1920's.

[Bunny_boy]

God, it's probably been said already, but ceding production of an art form to a transnational corporation doesn't fit into the definition of punk that works for me.

[Bombadil]

Is this the 5 minute argument or the whole ½ hour?
No one here is going to change anyone's mind about it.

All I know is:

I will not use this shite to generate 'original music.' Ever.
I will not listen to anything that has anything to do with generative AI.
There's enough real music out there, to my taste, from Bach to Pink Floyd, to keep me sated.

[jancivil]

I have no illusions on persuading anybody on the internet. (Talking to people I can be persuasive.) If I'm typing there'll be a couple basic reasons: there may be some educational value or thought it spurs, I just like writing it, or I'm venting. Guess which this is. Enough from this dipshit.

Here's the thing, if that guy was able to express himself and the AI has been trained on it, then the AI will be equally capable of expressing those same emotions when you ask it to. That's how it works.

No AI has been trained on my music, never will be. You have no clue about most music that's ever been made, that's abundantly clear, in the first place. Your view of things is about as narrow as can be.

We know you, you've not been obscure as to why you're in music. You have anger issues and you want to "express" that. Before you encountered this in music you weren't interested. You aren't interested in music, you understand fuckall about it. The statement "clicking notes in in a piano roll" is equivalent to having a machine come up with notes is easily the most ignorant thing I've ever seen in my life. One will have to explain this to you as though you're 5?

I have my own ideas. AI has no ideas. This is so basic, honestly, to fail to grasp it means literally unbelievable ignorance as to the act of inventing music or one is happy to look like an idiot just to make the idiotic point, so the move of a total antisocial asshole.

I am able to make my ideas happen a number of ways without much of a struggle. I don't, and won't know if an AI is able to somewhat ape things it finds some prevalence of trained on my music as it would be a massive waste of time. My music is not reducible like that. This thing may be so dissimilar to the last thing, stylistically, in terms of materials... the instrumentation inside of one piece might represent two wholly different bands (in one case four)... One thing might use conventional means, another is entirely off the wall. Some of those sounds never appear again.
in one case the microtonal aspect is not going to happen again, it stands alone to do this one thing one time.

AI might get confused. There is no simple emotion, I do music. I did a show once where because of the political content or point - at the War Memorial Opera House - I did mean to tug on the old heartstrings by the end (I can sang country) but we did other things on that journey.

Music to me, to a serious musician is not reducible to a need to express something. It's music qua music.
If you have a problem with someone 'clicking in notes in the piano roll', great let's have more of how you have no clue. I've done that and written entire pieces from a hospital bed. I got a hair up with an idea, it will have been a fleeting idea/spur of the moment so I set to work. I don't do it because I can't play, I was a performance major at a big conservatory once upon a time, an A grade from the toughest jury imaginable. I have big chops 'clicking in notes'.

You lack the very first clue as to what "AI" is or does. It has no emotional history, it has no thought, it has no capacity whatever to reason; it generally searches the internet with the intent of locating prevalence; technically it's an averaging engine on top of an insanely fast search engine. You want to believe it's expressing emotions, it isn't. It has literally no concept of expression. A belief that statement of yours is true is a delusion.

[Mike777]

Two recent productions of AI music I heard recently on YT were excellent from a musical and emotional aspect. The first was a trained classical musician give 5 or 6 audio clips of classical music, and asked to determine if they were from real composers are AI generated. He got two wrong I think and thought he was hearing a human classical composer. And thought the real human composer clips were AI.

Suno creating classical music is not a surprise. The 2nd thing I head blew me away.

Suno is trained very well in music production. I guy had improvised an original, improvised ballad playing his guitar and singing (badly) 30 years ago and there was one take. So it had acoustic guitar, his vocal, and lyrics. It was an ok ballad. Boring. But it did have real human emotion.

All he had Suno d was produce it much like it would be produced by real pro musician's and a pro singer. Suno added two elect guitar solos I think. The result of this average ballad became what I would say is a near masterpiece of song craft.

You could spend many hours producing/mastering your songs, and get good results. Even professional results sometimes. THE most difficult aspect of a song is the quality of the singer/vocals. Suno not only created a pro production from the 30 year old take of his ballad, but also produced a pro male singer. And because the guy's take had emotion, so did Suno's result. The emotional intent was very human sounding and not artificial.

[VOODOO U]

I literally can't see any similarities between choosing which AI generated audio from a single prompt you prefer to sitting with an experienced guitar player and trying different ideas (

You are 100% correct. And do you know what the key difference is? I can tell A.I. what to do over and over and over again. Again and again and again.
And it won't get pissed off, it won't be insulted that i didn't like it's first take, it doesn't need a break, and more importantly it will be available when I'm good and ready. Try that with some experienced guitar player.

It's dumb

Do the following equation in seconds: 2,673 + 4,654 + 5,212 x 3 + 8,789.
Ok times up. You failed. You're dumb.
But guess what? A calculator can do it.