Can AI (Artificial Intelligence) even compete?

[vurt]

]

This is false. All music, comes from "the past." Show me some music that can't be traced back to something that came before.

forbidden planet, the krell music.

im not even sure of any electro akoustiche stuff similar, pre fp?

[Bombadil]

^^^ it isn’t that I give AI too much credit; rather, I don’t underestimate human stupidity, avarice, or malice. AFAIK, any meaningful and impactful technological discovery has fallen into the hands of those who use it to exploit and divide…us.
Ape descendents fascinated by digital watches.
Humanity has a bad habit of doing things before asking whether they should be done.

[vurt]

Ape descendents fascinated by digital watches.

tbf to us, that does seem to be a primate thing, fascinated by tech, ive seen orangutan swiping on ipads. id hate to see what bonobos use them for

[Bombadil]

Great Apes have been taught to use computers to communicate with us. Quite fascinating, really.

Bonobos...well, they better be the age of majority in your country.

[vurt]

Great Apes have been taught to use computers to communicate with us. Quite fascinating, really.

and to write shakespeare for some reason, use a bloody photocopier!

[zerocrossing]

]

This is false. All music, comes from "the past." Show me some music that can't be traced back to something that came before.

forbidden planet, the krell music.

im not even sure of any electro akoustiche stuff similar, pre fp?

Right. A lot of "experimental" music is very commonly known as sci-fi or horror soundtrack music. I found I had a much better audience reception for my experimental and ambient stuff if I ran a movie along side it.

[Lunarkittn]

Mmmmh, so for you it's like this: all rock and pop music,
electronic music, ambient, etc. - starting with the Beatles - is
not necessarily "trash", but only "does not demonstrate a high
level of cultural expression".

Not starting with the beatles, I would go with earlier than ragtime even, but I'm not a historian.

Let's go back to what I said.

To me it feels like AI is pointing out how the production of music has taken turns in a direction, that arguably might not have been desireable from the get go. Historically music has been an expression of local culture or exceptional genius, as an expression of a higher culture. Popular music in contrast has swallowed alot of opportunities for local culture to exist to begin with. Why go and listen to local musicians when you can easily overstimulate yourself with music, video and all the other things your computer, TV or smartphone offers you 24/7 on demand. And even if you want to go out and experience a live performance, you might go for a concert from the usual suspects of the music industry instead. Not to speak of every other metal band that 'wants to sound like slipknot' or whatever. I'm writing this with the understanding that nothing that popular culture has produced is an expression of high culutre, including the beatles or what ever your favorites are.

Historically we might have had people sitting in caves making percussive sounds and singing in some shape or form. We had local cultures making their form of local music like (non-commercial) folk music for example. This was motivated by individuals in a local context.

Another thing we had is elites of different cultures such as mesopotamia, greko-roman antiquity, christian culture of the dark ages, the enlightenment and what followed, like Weimar Classicism. They enabled structures to form that create high culture in order to project power and structure.

When it comes to the beatles for example I'm a bit confilicted on what we are dealing with. I once spoke to someone who lived at the time of beatlemania. I told them that I found the phenomenon of beatlemania quite ridiculous and hard to believe that it was a real thing. They told me that it was part of the Centrl Intelligence Agenncy M K - Ultra project, which would give 'Lucy in the sky with diamonds' a whole other meaning. I find this to be quite a claim, that is asking for proof. Apparently there are people believing this, but without any real evidence I'm not willing to make that leap of faith. The conventional understanding is that the beatles is a capitalisitc commerical product. Regardless of what you believe, I'd say that music of such a project or of capitalist commercial intrest is an expression of low cultute. I hope this clears up what I meant to say.

Anyone who pursues this music of "low cultural expression"
or composes similar music themselves is not making a
relevant contribution to culture anyway. Therefore, AI poses no
danger in this regard.

If you're making music that is easily replaced by AI, I think it puts into question the worth of your work and that is not a loss in terms of cultural developement, yes.

So one should look within oneself: There is nothing there that could
be considered culturally valuable or interesting, that is "of high
cultural expression".

So, one should rather look at oneself and see if one could
contribute something "relevant" to the culture.

If they want to make something that is culturally relevant, yes.

I think I understood you very well. But I'm a little shocked.

You're evaluating people's original artistic expression and
dividing it into "important + valuable" and "less relevant +
not worth noting." That is, you're dividing art into "bad"
and "good."

I would go even further and say that if your work can replaced by AI, it is not even art.

While one can certainly do so according to personal taste, such
a distinction is of little use as a basis for objective discourse. In
art itself, such a distinction is even highly problematic — indeed,
absurd.

No, if your work is replacable by a mimic mashine it is objectivly not art. Art needs to be of a quality that can't be just replicated by anyone or anything. A nail is absolute genius, but it's easily reproduced in a way that the original adds nothing extra. If there was only one nail and no one and nothing could replicate it's qualities it would be a piece of art. But it isn't and therefore a nail by itself is objectivly not art, as is your music if it is replaceable by AI.

Because art in general, and music in particular, CANNOT be
evaluated. What, then, is Beuys' Fat Chair? Or Malevich's Black
Square?

I don't know these examples. By the name of the 'Black Square' it sounds like an expression of a saturnian cult.

Overall, you are completely wrong in two respects, or rather, you
are even deceiving yourself:

1. In the AI ​​discussion, you are escaping the real issue by
arbitrarily dividing music into "bad and not worth considering"
and "culturally valuable".

Demonstrate how this is wrong.

2. And you completely suppress and negate the idea that in the
future it will be easy for AI to compose music in the style of
Bach, Beethoven, Debussy and others, so that even experts
would be amazed (see my point 4 from the first post).

They may compose in the style of classical composers, but they can never compose like a classical composer, since their work is an expression of human experience and an AI is a mimic mashine without human experience period.

[Opaque]

They may compose in the style of classical composers, but they can never compose like a classical composer, since their work is an expression of human experience and an AI is a mimic mashine without human experience period.

This statement counters everything you've written above, since the same could be applied to any of the works you'd probably characterize as 'low culture'.

They may produce in the style of Kendrick, but they can never write like Kendrick, since his work is an expression of human experience and an AI is a mimic machine without human experience period.

[Lunarkittn]

They may compose in the style of classical composers, but they can never compose like a classical composer, since their work is an expression of human experience and an AI is a mimic mashine without human experience period.

This statement counters everything you've written above, since the same could be applied to any of the works you'd probably characterize as 'low culture'.

I don't see how this statement not exclusivly applying to high culture counters anything I said.

They may produce in the style of Kendrick, but they can never write like Kendrick, since his work is an expression of human experience and an AI is a mimic machine without human experience period.

Agreed

[Opaque]

Sure, but then the question of one's work being replaceable by AI is not about outwards impression, because that depends on people's perception of the difference or similarity between the two, as well as their familiarity with the genre/style/tradition, etc., but is more focused on the source and process - human experience, motivation, etc.

In other words, statements like "I would go even further and say that if your work can replaced by AI, it is not even art" become problematic because the question is who is that authority that can determine such a thing, especially when such strong statements about art come from someone who has never heard of Malevich (so probably not very familiar with 20th century art or well read on modernism, art history, and so on).

[enroe]

...

In other words, statements like "I would go even further and say that if your work can replaced by AI, it is not even art" become problematic because the question is who is that authority that can determine such a thing, especially when such strong statements about art come from someone who has never heard of Malevich (so probably not very familiar with 20th century art or well read on modernism, art history, and so on).

Yes, that's exactly what I think too.

[enroe]

First of all, thank you, Lunarkittn. You really took the time to
answer in detail. Nevertheless, we have very different opinions.

----- Example "Beatles" -----

You say the Beatles are "low culture" because they are a
commercial capitalist product.

That's somewhat endearing, because the commercial behemoth
simply devours so much and destroys so much. But when it comes
to evaluating culture as "inferior" or "high-quality," we need to take
a different approach:

Whether a work of art is "high-quality" and thus contributes to culture
is decided right at the beginning; indeed, it is even decided by
the artist himself. According to Joseph Beuys, everyone is an artist
if they create a truly original work of art.

Only then do the companies come along and say: We'll market this
and that, and not that other thing. But just because they market "this
and that" doesn't make "high-quality art" "inferior art." The
"marketing" itself might be considered disreputable or even, in
some cases, "destructive," but it cannot devalue the art itself.

Therefore, for me, the Beatles, the Nobel laureate Bob Dylan,
Genesis and many, many other "high art" remain, with a fantastic
contribution to human culture.

The economic marketing makes this part of the culture audible
to many people - and that is also a great advantage.

----- Summary -----

Overall, you are completely wrong in two respects, or rather, you
are even deceiving yourself:

1. In the AI ​​discussion, you are escaping the real issue by
arbitrarily dividing music into "bad and not worth considering"
and "culturally valuable".

Demonstrate how this is wrong.

2. And you completely suppress and negate the idea that in the
future it will be easy for AI to compose music in the style of
Bach, Beethoven, Debussy and others, so that even experts
would be amazed (see my point 4 from the first post).

They may compose in the style of classical composers, but they can never compose like a classical composer, since their work is an expression of human experience and an AI is a mimic mashine without human experience period.

No, no, that's completely wrong.

I think you, and probably many others here, haven't quite
understood what the new, emerging AIs really are - and what the
difference is:

AI is not a "bigger computer" or a "bigger program" with a
"mega-sized database."

No! AI is a "neural network" that can learn on its own. Much
like a small child, only a million times faster—and with a million
times more data. And that means we are entering a new era:
Many demanding professions, such as those in law firms and
engineering offices, will be replaced. An employee of OpenAI
once described AI as the "atomic bomb of the 21st century" —
that sums it up quite well.

And that also means: The AI ​​will compose new music in ALL
genres. And in the highest quality – in the style of any specified
musician, whether Phil Collins or Johann Sebastian Bach.

This is a problem for creative musicians, because their (and
my) creativity and music are simply pushed to the wall and
wiped out by this overpowering AI competition.

[Zeisner]

No! AI is a "neural network" that can learn on its own.

Nope. It needs to be trained, conditioned.

Much like a small child

No because small children can think, they can reevaluate. Like "How many fingers does a human have really?". AI is not able to do that, it's just a stochastic parrot.

only a million times faster—and with a million times more data.

Which is pointless considering the low quality of the training data (Like audio processed with lossy compression codecs) which leads to artifacts. AI is not able to fill in those gaps because it sucks at interpolating and extrapolating because it can't think nor sense. It is not able to understand how broken the dynamics, frequency spectrums and stereo fields of its training data really is. Nor can it understand how perspective can make a hand look like as if it only had four fingers. It can't think in concepts nor can it think outside of the box. It's just a stochastic parrot.

And that also means: The AI ​​will compose new music in ALL
genres. And in the highest quality

AI already hit a plateau. It doesn't get better than that. It would require a lot more (and better) training data, a lot more data centers and a lot of experts to finetune the model - top musicians and audio engineers from all around the world, the entire upper quartile. The costs would be insanely high. Take all money previously spent on AI, multiply it with 1000 and you're not even close. It's more realistic to invent the warp drive.

[jancivil]

And that also means: The AI ​​will compose new music in ALL
genres. And in the highest quality

To say this looks like one is operating from an understanding of music that's seriously thin and shallow at best. It doesn't comport at all with an understanding of any but a seriously limited process of creating music.

You've given us a litany of unsupportable beliefs about this technology that would all of it be disabused by a little bit of research. There is a point past where the term for this is a delusion, and the above assertion (CF: all caps "ALL") shows that rubicon well crossed.

It seems you're entirely impervious to everything that doesn't prop up the belief.

A machine with no awareness or ability to experience anything of life having no possibility to have sense of qualia whatsoever, you consider capable of jazz; now it's going to be "in the highest quality". It does not, cannot think. It has absolutely no capacity to discern quality. Ask someone in the business of developing the Large Language Model. It does pattern matching: improvisation demands discernment as per 'which pattern' to even begin to participate. It's little hand will need to held and it guided at every point, IE: told which pattern.

Here's an hilarious example. An LLM was given an arithmetic problem, the kind in grade school known as a 'word problem' (a type of reasoning problem is embedded within it) as opposed to what you'd enter into a calculator.
"Oliver picks 44 kiwis on Friday, then he picks 58 kiwis on Saturday. On Sunday he picks twice the number from Friday, but five of those are smaller. How many kiwis has Oliver now?

The thing does 44 + 58 = 102; 44 x 2 = 88; 102 + 88 = 190; minus 5 = 185.

We cannot draw from this that it misunderstands the word smaller; it doesn't know how to think, it has to be told. It had no clue "smaller" was extraneous to the problem. There is no logic to explain 5 kiwis ceasing to exist in the language provided.



Took me the time to type "artificial intelligence limitations", hit enter, click on the first hit.

"LLMs struggle with logical reasoning because of something called token bias." (tiny tweaks in the input having an outsized result in the output).

The plan with extant models is more sophisticated training such as to induce "train of thought" in the prompting as well as "inference time".

The key difference between "Chat GPT" and AI generating musical product is known as iterative vs generative. The former is basically conversational, the latter is a command.
It's a command in verbal language nonetheless.

Jazz improvisation has no time for the latter, and again LLM will need constant prompting from verbal language. "Generative AI uses algorithms trained on large datasets to learn patterns to create new content that mimics the style and characteristics of the original data."

Capable of mimicry doesn't through itself translate to an ability to communicate thoughts as a genuine responses. In the LLM of either type its dependence on prediction via statistics is absolute.

No jazz improvisor is reliant on mimicry as a product of scanning an abstraction from a large database. There is no time, you have to understand right the fvck now what is happening and be genuinely responsive. The thinking has to be ultra focused, I don't even like the word thinking for it.
It's not about spitting out copies found to be of highest prevalece to a verbal prompt. You have literally no concept of the matter.

[enroe]

No! AI is a "neural network" that can learn on its own.

Nope. It needs to be trained, conditioned.

And that also means: The AI ​​will compose new music in ALL
genres. And in the highest quality

To say this looks like one is operating from an understanding of music that's seriously thin and shallow at best. It doesn't comport at all with an understanding of any but a seriously limited process of creating music.

Mmm ... we've been through all this before:

Of course, you can search the internet for "your opinion" and then list it
here. But that doesn't change the facts or the future prospects at all.

But ... and this is what's happening to you here: If you lie to yourself,
saying "AI could never do that" or "AI is far too limited for genuine
expression," then this has a psychologically soothing and comforting
effect, calming and reconciling your inner child. You lie to yourself: "AI
can never replace creative people," and suddenly the world is right
again.

If you need this – and if it's important to you – then
stop here! Don't read any further!

-----------------------------------------------------------------------------------------------

I already distinguished between these 3 categories at the beginning:

1. AI Tools + Plugins
2. AI Song Composition - in 1-3 years
3. AI Song Composition - in 4 years and beyond

We will now look at category (3). To do this, we need to extrapolate a
bit: What will AI music creators look like in 4 years? How long are the
development cycles? How long does it take to learn, adapt, and gain
market access?

As I mentioned before, AI is not a large computer program. Although
both Zeisner and Jancivil believe exactly that and demonstrate it here.
AI has a completely different foundation – namely the concept of "neural
networks" – and is equipped with "learning capability." This means that
AI can learn on its own like a child, only a million times faster, and with
a million times more data.

To understand this, you need to know: From a computer science
perspective, the human brain is nothing more than a neural network,
which functions according to the same physical rules. AI can do all of
this now too, only a million times faster.

Entire armies of university graduates (computer scientists,
mathematicians, physicists) are further developing neural AIs and
optimizing their learning environment. All of this is being pushed with
billions of dollars (!), meaning the development is almost unpredictable.
It will completely revolutionize our everyday lives. When someone
contacts their health insurance company, they will first speak with the AI.

And that brings us to (3): The successors of SUNO, UDIO, and others
will no longer have the initial teething problems. They will play on a
completely new level, both musically and in terms of sound. And that
naturally means:

The AI ​​will compose new music in ALL genres. And in the highest quality
– in the style of any specified musician, whether Phil Collins or Johann
Sebastian Bach.

And this has consequences:

The big companies and those dominating Spotify have discovered:
With AI creators, they can generate songs in no time, upload them
in droves, and even make money. Companies are even using AI to
design live groups, partly human, partly avatar, which are presented
as live performers and deliver fantastic, breathtaking, even superhuman
live performances. TV and internet platforms will be flooded with
them — teenagers are intoxicated by them and click on them.

But it won't be that bad:

An older participant at a meeting raises his hand: "Where do people
still actually sing and play guitar?"

A voice from the crowd replies: "You can do that at the senior citizens'
center. My grandma goes there too. They still do such strange things!"