A.I. "Jazz"

[MathOfInsects]

If someone's music is truly in danger of being replaced by AI, then AI isn't the problem.

[ProfD]

10 hours ago, CyberGene said:

I hope this doesn’t get interpreted as disrespectful but IMO jazz is the easiest task for a proper generative model. Jazz is so formalized with all possible devices of improvisation, (re)harmonization, voicing and rhythm structures described and taught to death in any possible online or academia courses, that it’s actually a wonder to me why it hasn’t been done earlier. 

There is a very real reason that no amount of academia is capable of stamping out another Herbie Hancock or Chick Corea.

 

10 hours ago, CyberGene said:

...making an acceptable jazz improvisation is a matter of practicing enough, it’s not a divine stroke. Perfect task for an automat. 

False.  More than likely, the end result will be generic and nothing worthy of consumption.

 

Creativity and orginality in music is divine.

 

AI will produce music in the same way that a calculator solves math problems. Logical.

 

Playing and producing music beyond rote requires ear and feel; things that cannot be taught to another human being or data fed into a computer.

 

When someone is enamored with what a musician plays, it has more to do with their approach to the music more than the notes they're actually playing.   

 

The bottom line is...a computer will never be able to produce music from feeling it.😎

[Docbop]

10 hours ago, CyberGene said:

making an acceptable jazz improvisation is a matter of practicing enough, it’s not a divine stroke. Perfect task for an automat. 

 

No not for REAL Jazz.      Too many people Jazz is just a style and rolling their eighth notes and adding a 9th to a chord makes them a Jazz player.   Jazz is an approach to playing, I think it was Pat Metheny that said Jazz is a verb.    

[CyberGene]

1 hour ago, ProfD said:

There is a very real reason that no amount of academia is capable of stamping out another Herbie Hancock or Chick Corea.

Well, that’s what I said in the same post. AI can’t come up with a new Herbie. However, it can certainly generate jazz improvisations that will be more or less good enough for anything less than Herbie 😀 I don’t understand what’s bad in saying it loud though? I don’t underestimate jazz players. Instead, I believe AI is (or will soon be) good enough to be a great improviser. 
 

1 hour ago, ProfD said:

More than likely, the end result will be generic and nothing worth of consumption.

Yeah, because a lot of jazz is nothing like that 😉 I said “acceptable”. I find many human made improvisations even less worthy of the label “acceptable”. And even some by famous jazz players. Let’s not pretend that all famous jazz musicians have produced only stellar impros all the time.

 

1 hour ago, ProfD said:

Creativity and orginality in music is divine.

Yes. Again, only a few of the most famous classical composers and jazz musicians are in that category. Everything else is just a bit less divine 😉 IMO 😀

 

And once again, what I’m saying is, AI will be good enough to pass as “great”. Maybe just not divine. That’s what I said, just re-read it. 

[CyberGene]

P.S. Jazz is an improvisation skill that uses a particular music language. The latter is formalized and a lot of it is in books and courses. Whereas the skill to do it in real time is both a gift and a result of a lot of practice, good ear and quick instincts (which in turn is just brain power). Apart from arguments about what constitutes gift, the rest is just a bunch of almost mathematical calculations performed in real time. Great task for a computer (regardless of whether it’s AI-based or pre-programmed algorithms. The advantage of AI is just the ability for algorithms to adapt and transform into producing particular desired output, based on training with given input and evaluation weights/scores). 

[CHarrell]

Any AI algorithm trying to reproduce an improvisation like on a solo, would have to have enough randomization and variation to respond to the countless stimuli that can influence a musician's playing, such as a drummer responding to a pianist's phrase by mimicking it in a call-and-response fashion, and the pianist potentially continuing that in kind, a pianist playing a certain kind of harmony while comping, inspiring the horn player to play some different notes (or vice versa, maybe a horn player plays an outside line or note inspiring the comping pianist to change up their harmonies to lean into it), or a certain phrase from a player inspiring the dynamics of the group to shift...these are some very small, linear examples of that kind of real-time thinking and adapting. Which of course, as we know from scary self-driving car footage, seems to be one of AI's greatest weaknesses. 

[CHarrell]

5 minutes ago, CyberGene said:

The advantage of AI is just the ability for algorithms to adapt and transform into producing particular desired output, based on training with given input and evaluation weights/scores). 

 

How funny you say that as I typed out the opposite! 😂 

[CyberGene]

Self-driving car AI is a special case which involves safety and regulations and is thus severely limited in its scope of action. BTW, I can assure you, current self-driving AI will most probably drive better than most people, and will result in much less casualties, despite occasional weirdness. So, it’s not a divine driver but is nevertheless a great driver that is better than 99% of the drivers 😀

[ProfD]

14 minutes ago, CyberGene said:

P.S. Jazz is an improvisation skill that uses a particular music language. The latter is formalized and a lot of it is in books and courses.

 

Apart from arguments about what constitutes gift, the rest is just a bunch of almost mathematical calculations performed in real time. 

Rest assured that a musician who *thinks* or believes or approaches Jazz or music in general from that standpoint will miss the mark every time.

 

16 minutes ago, CHarrell said:

Any AI algorithm trying to reproduce an improvisation like on a solo, would have to have enough randomization and variation to respond to the countless stimuli that can influence a musician's playing...

Exactly. 

 

Jazz musicians decide to play outside and come back inside and the interplay between them is impossible to turn into an algorithm.  Especially when so much of it happens in the moment...improvisation.😎

[CyberGene]

5 minutes ago, ProfD said:

Jazz musicians decide to play outside and come back inside and the interplay between them is impossible to turn into an algorithm.  Especially when so much of it happens in the moment...improvisation.😎

Everything can be turned into an algorithm 😀 Spontaneous behavior is either random or triggered by stimuli. Both are features that are available to computers.

 

Being able to react to other musician playing at the same time is what? Paranormal activity? 😉 What’s preventing the computer from listening to the band in realtime and reacting the same way as a human? 

[ProfD]

13 minutes ago, CyberGene said:

Everything can be turned into an algorithm 😀 Spontaneous behavior is either random or triggered by stimuli. Both are features that are available to computers.

 

Being able to react to other musician playing at the same time is what? Paranormal activity? 😉 What’s preventing the computer from listening to the band in realtime and reacting the same way as a human? 

I'm confident that I'll never walk or be rolled into a Jazz club or any other venue and see robots or holographs playing real Jazz. 🤣😎

[CyberGene]

See, a car is faster than you can run. But we still run. Occasionally 😉

 

Fear of technology has been a thing since the Industrial Revolution. We’re still here and kicking. As someone else said above, there’s no money in jazz, so AI is moving on to more lucrative genres anyway 😀

[ProfD]

1 hour ago, CyberGene said:

Fear of technology has been a thing since the Industrial Revolution. We’re still here and kicking. 

To be perfectly clear, I have zero fear of technology. 

 

I fully embrace technological changes and process improvements especially when it efficiently makes life easier.

 

That's why I'm neither a vintage gear purist nor a hardware or software *hater*.  Use the best tool for the job. 

 

AI will be a useful tool.  No more or less.😎

[CyberGene]

And to clarify, I don’t think AI is there yet to create any listenable jazz. But it can certainly do soon, taking in mind how much AI is becoming a reality. And provided there’s interest in programming it to play jazz at all.

 

I’m not an atheist. However, I’m not religious either and I don’t buy the idea of how art and creativity is something coming from some metaphysical space, divine, etc. It’s all a product of great minds that happened to be optimized for that particular field and be at the highest possible peak level on a world scale. Can this be recreated through machinery? Most probably. Can we humans create that machinery/models? Not sure. Or maybe not soon. 

[jazzpiano88]

3 hours ago, ProfD said:

The bottom line is...a computer will never be able to produce music from feeling it.

 

As a devil's advocate, some believe AI will reach sentience.    One guy got fired from Google for thinking, and broadcasting, it already has. 

I vehemently disagree since I've done it, and it's just a Python program executing a series of weights through matrix multiplies on a GPU.

 

 

[stoken6]

I can see AI Jazz (at least a generative pre-trained form of it, incorporating some of the convolution-like feature detection I mentioned earlier) being able to achieve in the future what ChatGPT and the like can do today with natural language. It will be derivative, not innovative, but it will be passable.

 

We are not there today.

 

Cheers, Mike.

[AROIOS]

21 hours ago, CyberGene said:

... jazz is the easiest task for a proper generative model. Jazz is so formalized with all possible devices of improvisation, (re)harmonization, voicing and rhythm structures described and taught to death in any possible online or academia courses, that it’s actually a wonder to me why it hasn’t been done earlier...

Yup, PG Music already attempted it 25 years ago with their "Soloist" and "Melodist" features in Band-In-A-Box. 95% of what it generated were insipid and sometimes outright jarring results. But for a songwriter, the 5% usable ideas more than make up for the fluff.

As I've said before, improvisation is mostly just a form of mental masturbation, enjoyed only by the performer and a few spectators. To most casual observers, it bears little difference from the output of a computer arpeggiator with a bit of pre-programmed randomness. Human intelligence is vastly more powerful than those simple neural-logical algorithms that churn out "Jazz Improvisation".

But we artists are a notoriously self-important bunch, so thumbs-down from the non-techies among us is hardly surprising. 😃

[AROIOS]

12 hours ago, MathOfInsects said:

If someone's music is truly in danger of being replaced by AI, then AI isn't the problem.

Until A.I. passes a Turing's test for the particular musical task it's asked to fulfill.

In other words, as soon as A.I.-generated content become indistinguishable from that from a human counterpart, at similar cost, Mr./Ms. human becomes "useless" for that task.

[AROIOS]

8 hours ago, stoken6 said:

I can see AI Jazz (at least a generative pre-trained form of it, incorporating some of the convolution-like feature detection I mentioned earlier) being able to achieve in the future what ChatGPT and the like can do today with natural language. It will be derivative, not innovative, but it will be passable.

 

We are not there today.

 

Cheers, Mike.

I suspect artistic "creativity/innovation" is little more than 'hallucinations" displayed by the LLMs and Multi-Modal Models right now. Maybe they are just "tasteful" random mutation that happened to tickle us "the right way" and dwarf in comparison to "creativity/innovation" in STEM, the kind that brought us Calculus and Relativity.

[AROIOS]

12 hours ago, ProfD said:

...
Creativity and orginality in music is divine.

...

 

10 hours ago, CyberGene said:

Yes. Again, only a few of the most famous classical composers and jazz musicians are in that category. Everything else is just a bit less divine 😉 IMO 😀

 

And once again, what I’m saying is, AI will be good enough to pass as “great”. Maybe just not divine. That’s what I said, just re-read it. 

Yup, "originality" only describes 0.01% of what happens in music or any artistic pursuit.

The rest 99.99%, are just regurgitation and imitation.

[AROIOS]

16 hours ago, nursers said:

 

[Verse] Come on down to The Keyboard Corner Where the ivories play all day and night (ooh-yeah) A forum for players, beginners and pros We gather 'round, sharing our delight

 

[Chorus] At The Keyboard Corner, we find our melody From bebop to swing, improvising in harmony Together we create, a magical sound At The Keyboard Corner, where jazz is found (ooh-yeah)

 

 

This is showing potential. Give it a few years and the stock music industry will be f***ed like ProfD and CHarrell mentioned.

[AROIOS]

8 hours ago, jazzpiano88 said:

...it's just a Python program executing a series of weights through matrix multiplies on a GPU...

 

How do we know human brain isn't just that, except scaled by 100000X? 😃

Even some of my friends in High Performance Computing throw around comments about the "divine human ingenuity" with wild abandon.

I just find it amusing how little difference there is between their blind faith in the human mind, and the religious fervor of rednecks they look down upon. 😆

[AROIOS]

9 hours ago, CyberGene said:

...I don’t buy the idea of how art and creativity is something coming from some metaphysical space, divine, etc...

Yup, and you can't have an informed conversation about A.I. with folks who don't even understand Gradient Descent and believe "if/then statements" are all there is to it.

[jazzpiano88]

2 hours ago, AROIOS said:

11 hours ago, jazzpiano88 said:

...it's just a Python program executing a series of weights through matrix multiplies on a GPU...

How do we know human brain isn't just that, except scaled by 100000X? 😃

That’s a big part of the problem.  We understand virtually zero about human consciousness.  Therefore we can’t prove or disprove that a rock is conscious and stuck in a rock’s body.  

[jazzpiano88]

15 hours ago, MathOfInsects said:

If someone's music is truly in danger of being replaced by AI, then AI isn't the problem.

True …. But…..  Thousands of people are replaced in their jobs on a daily basis by people (humans) who can do a job better than they.   Half the people (lower half of the bell curve) suck at their jobs comparatively and are continually being replaced and moving around from job to job.  AI will replace them first.  
 

[Thethirdapple]

RANT ALERT:

 

love it or hate it...

We choose to embark down a road of commodification without clear definition of who is who, not good...

 

By any other name?

 

PI - Pretend Intelligence

FI - Fake Intelligence

AI - Almost Intelligence

 

or any OTHER synonym...

 

unnatural, strained, mock, fake, false, mechanical, simulated, pseudo, affected, exaggerated, plastic, contrived, theatrical, forced, factitious, theatric, phony, spurious, assumed, bogus, cute, feigned, unreal, sham, hollow, automatic, pretended, empty, put-on, formal, manufactured

 

Still want some sociopathic algorithm to fulfill your every prompt ?

Oh, the marketing hurts so good...

 

WE are trained to value "natural" except when it comes to making decisions ?!?!

have fun with that..

 

Don't get me wrong, I love electricity and even enjoy integrated circuits and transistors, but we as humans have difficulty in knowing when is enough. To what end, for what purpose. Imagine inventing a hammer without anything to hit, that world would suck. As "its" being marketed to us, "it" serves NO purpose, fulfills no need, other than a revenue stream.

 

WE are not the consumer, we are the commodity

 

Not to even really mention the Ouroboros effect: the dataset is scraped up from what we create, then we use the dataset to "create" stuff which is created from the dataset we created from... might as well go back to being 10 years old, copying cassette tapes and wondering why the quality sucks more and more at each duplication of the duplication.

 

This thread along with a few others, combined, are being to mesh into an interesting fabric.

 

PEACE

_

_

_

[jazzpiano88]

One of the major things that sparked the AI, or Machine Learning, revolution was the plethora of training data, namely image data.   Large (Deep) Neural Nets capable of doing amazing things were not practical without enough data to train them without overfitting.  

 

 So with a large image data set available called imagenet the researchers toiled away with various network topologies, training techniques and hyper parameters to see who could achieve the best image classification accuracy on millions of images for object recognition. It was literally a completion. 
 

Along the way, they started using the networks to do other things, like generate continuous outputs to steer a car.  And here we are.  I prefer the term Machine Learning over AI because I think it’s more descriptive.  We’re not really mimicking “intelligence” IMO.  

[Thethirdapple]

4 hours ago, jazzpiano88 said:

Half the people (lower half of the bell curve) suck at their jobs comparatively and

And the other half of replaced people simply suck profit away from corporate greed …🫣

 

 

PEACE

_
_
_

[Reezekeys]

Great jazz improvisations - the kind that advance the art form or just illuminate a particular player's singular talent - often come from "mistakes", or breaking "rules." I would be interested in seeing how an AI can be programmed to do that. Introduce "randomness"? Maybe, but remember it's humans who will be the judge –  and there's over 100 years of history that includes listening to the greats like Armstrong, Ellington, Bird, Miles, Coltrane, Ornette, etc. I also suspect that their advancements, even if sparked by "mistakes", also came from methodical work and trying out new ideas. How does an AI decide which ideas "sound good" or not?

 

There's also the fact that many of these advancements come not from individual expression, but from a synergy of multiple players reacting to each other. The best jazz I can think of is that made from collective effort - Miles' two great quintets, Coltrane's classic quartet, Duke Ellington orchestra, etc. I wonder how that get coded into an AI program.

 

Even if an AI can someday paint a Mona Lisa or produce a credible Bill Evans piano track, by reason of how it was produced I maintain it cannot have the same value as the human-produced source it draws on. If anyone can click on a few buttons and get a Mona Lisa or a Coltrane soundalike track, what does that say about our capacity to honor or enjoy human creative work? It's gonna be a boring world. Am I imagining it or are some folks actually excited about this prospect?

 

Of course, in the end artistic value is in the eye and ear of the beholder. Given the dumbing down of popular culture, I have no doubt AI will be able to paint pictures that can hang in the walls of a Motel 6 or play muzak in a department store. Produce a credible Bill Evans improv? I very much doubt it. Let's say for the sake of argument that AI does indeed reach that milestone - I might listen to it out of curiosity, I might be impressed by what the AI  did, but the feeling I get from listening to the real Bill Evans would be entirely absent. And that "feeling" is pretty much the only reason I enjoy listening to jazz!

[MathOfInsects]

I would bet that generative AI would actually be superior to expanding/expounding on a mistake than people are. The options would be nearly endless, and would, based on past examples, almost certainly be justified-sounding and interesting. What it won't do is expose anything human about the "generator," and IMO that will forever remain true. 

 

The bottom line is, if we're able to be copied and replicated so frighteningly, we just have to up our game. The real reason people are scared isn't because of lofty societal reasons, IMO, but because generative AI has the potential to expose us as creatively lazy and predictable, and we've been able to make money and have careers built on some degree of sleight of hand. Generative AI is the product of our own recursivity. So we need to double down and be uncopyable. That's all upside IMO.