Has AI attempted one?

Anyhow, I find this to be one of the best attempts by AI in imitating classical music:

A Little Bach AI Music -- composed using MuseNet artificial intelligence by OpenAI

 

Mozart can be formulaic and predictable while being Mozart, but nobody can be Mozart.

 

I think of your comment while listening to the ubiquitous style of melody of fast repeated notes supported by (relatively) light-sounding harmony, of Die Zauberflöte.

 

I believe it will be able to mimic his style, but a great composer isn't just an amalgamation of their style. These AI tools build algorithms based on patterns found in the music it analyzes. That's not what a composer does when creating something. That will be enough to probably fool most people in a blind comparison but it takes longer to get better. Remember, this stuff is brand new and is already everywhere. I think ChatGPT just came out two years ago and is much better than it was then but isn't an expert at anything. Ask your question in five years from now and most people (including experts) won't be able to tell the difference between an obscure Mozart piece and an AI creation in that style.

 

Ok, this is actually a really deep question. I've not stopped thinking through this topic. There was an excellent NOVA episode about possibly finding a new painting by Leonardo daVinci. This would be a very, very big deal because there are less than 20 surviving paintings by him, and all are masterpieces. Frankly, I forgot the outcome of the investigation, but what was really interesting was how hard it is to know if it's a work truly by him. So think about this, he had proteges and students, who mimicked him so well, that it literally takes forensic experts to determine was it him or a student who did a painting. So, if a student did a painting that only a handful of people can tell isn't by the genius mentor, wouldn't that make them equivalent? The student works are considerably less valued though no one can really tell the difference because one is authentic and the other a mimic. Does the same apply to Jackson Pollock? What if AI created a work that had all the characteristics of Jackson Pollock, wouldn't that effectively be like the student of daVinci? Hardly noteworthy but looks exactly like the original. It's a complicated question because it ultimately comes down to what is art and what is creative genius?

 

No, the student is by no means equivalent. They would be equivalent if they developed their own vision and style all of their own that was unique, original and of high aesthetic value, and created works that came from their own imagination, not from someone else's.

When doing pastiche or copying someone's work, the original artist has already done all of the hard work for you.

 

In terms of the creative process there is obviously a world of difference between a great composer creating a masterpiece and an AI generating it by mining data and applying some randomness.

Why does the creative process matter? I think it matters because it encompasses the composer’s purpose in creating it, as well the composer’s unique personality, influences, life experience, and choices. It means the work comes from somewhere, is made using a set of skills and knowledge unique to that composer, and it has a reason to exist.

So I would argue that the creative process is the whole thing - the entire value of a work. The score and any performance or recording are merely records of that creative process, and evidence that it happened and what it was.

If we have now invented technology that can produce forged records of a creative process that didn’t happen, then I do not see how we could attribute any meaning or value to it. And, sadly, the prevalence of such forgeries would undermine the value of genuine creations - especially if these forgeries become indistinguishable from the real thing.

Think about it. If AI music was prevalent, who would listen to music any more? Why would they want to? As aural wallpaper to make sounds in the background? You simply couldn’t listen to it the way you listen to, say, Schubert’s lieder, because you’d know there was no reason for its creation and it doesn’t mean anything. So AI would be the death of classical music, except than that we could still listen to authentic historic works, and perhaps devise a way to authenticate new human works.

Of course you could argue that human composers create their work in exactly the same way as AI - they mine data (past works, musical theory, existing themes) and apply some randomness in choosing options as they go along. You could argue there is no meaningful difference between our intelligence and what we call artificial intelligence, and these are really two brains that work in equivalent ways. You could argue that AI possesses cognition in every way that a human does. I cannot disagree with this. But, still, I don’t think it is unreasonable for us to assign value exclusively to the achievements of our own species. And we still wouldn’t listen to AI music, for the same reason that we don’t generally bother to buy albums of whale song.

 

This is exactly the argument Stravinsky made against radio and can be used for any ubiquitous game changer technology like television, internet, iphone, AI.

"The propagation of music by mechanical means and the broadcasting of music - that represent formidable scientific conquests, which are very likely to spread even more - merit close examination as for their importance and their effects in the domain of music. Of course, the possibility for both authors and performers to reach the masses, and the fact that these masses are able to make themselves acquainted with musical works, represent an unquestionable advantage. However, it cannot be concealed that this advantage is dangerous at the same time. In the past, someone like Johann-Sebastian Bach had to walk ten leagues in order to hear Buxtehude perform his works. Today, any inhabitant of any country simply has to either turn a knob or play a record in order to listen to the piece of his choice. Well! It is in this very incredible easiness, in this very lack of effort that lies the vice of that so-called progress. In music, more than in any other branch of art, comprehension is only given to those who actively contribute to it. In itself, the massive reception is not enough. The listening of certain combinations of sounds, and the automatic growing accustomed to them does not necessarily involve the fact of hearing and grasping them, for one can listen without hearing, the same way one can watch without seeing. What renders people lazy is their lack of active effort and their developing of a liking for this easiness. People no longer need to move about as Bach had to; the radio spares them the traveling. Neither do they absolutely need to make music themselves and to waste time studying an instrument in order to know the musical literature. The radio and the disc take over. As a result, the active faculties, without which music cannot be assimilated, gradually atrophy among the listeners who no longer train them. This gradual paralysis leads to extremely serious consequences. Overwhelmed with sounds, the most varied combinations of which leave them indifferent, people fall into a sort of mindless state, that deprives them of all ability to judge, and renders them indifferent to the very quality of what they are served. In the near future, such disorganized overfeeding is more than likely to make listeners lose their hunger and their liking for music. Indeed, there will always be some exceptions - some people within the hoard will be able to select what they like. However, concerning the masses, one has all the reasons to fear that instead of generating love for and understanding of music, the modern means involved in spreading music will lead absolutely to opposite results; it is to say, they will lead to indifference, to the inability to recognize them, to be guided by them, and to have any reaction of some value."

Igor Stravinsky - "Chronicles of My Life" – 1935

 

the reason is that you cant teach musical genius, you cant learn musical genius, and you sure as hell cant buy it on Amazon.com.

you either have it or you dont

I know many music theory professors that cant play thier way out of a paper bag even though they have studied many scores and know music theory.

there is actually more to it than that

 

Wasn't there a project to use AI to complete Mozart's Requiem? What happened to that?

 

It got bored and wrote a new Willie Nelson album instead.

 

I sure hope these genius programmers are working on inventing AI orchestras, so nobody has to learn to play an instrument anymore. If they need a bunch of money for such a task, they should build an Artificial Audience of robots to attend the live concerts first, then use the money for the computer orchestras. Who knows, maybe we won't have to be bothered with music at all someday and the computers can perform and consume for each other on our behalf?

 

AI can't create authentic art, period.