If A.I. took the place of musicians, would you listen to it?


A few questions which I'm curious about. If you have a take on this, please share!

Here's the question:

A.I. is increasingly playing a role in music creation. Not just assisting composers, but generating music.

If you found an A.I. generated song to be enjoyable, interesting, etc. would you have any objection to supporting it by listening and paying for the service which provides it?

If more and more music was like this, and there were fewer and fewer jobs for musicians, would that bother you? -- I'm thinking here about the aesthetics of the issue, not the economics or justice of it. 

I'm trying to understand if people just want to have a certain set of sensations from music and they don't care if there are human beings creating it -- or if it's important for you to know that what you're experiencing from music (or art) is coming from human beings.

Thank you for thinking about this.

128x128hilde45

First we are confronted with a semantic difficulty which mirrors our cognitive difficulty.

artificial /är″tə-fĭsh′əl/
 

adjective

  1. Made by humans, especially in imitation of something natural.
    "an artificial sweetener that replaces sugar; artificial flowers."
  2. Not arising from natural or necessary causes; contrived or arbitrary.
  3. Affected or insincere.

The British mathematician Alan Turing posed a test (the Turing Test) to determine whether a computer was as smart as a human. It was simply that if you interact with a computer and you can't tell if it is a computer or a human being responding then the computer was is as smart as a human. In the case of music, it you can't tell the difference between artificially generated music and human generated music then the artificially generated music is as good as that created by a human.

A more general question is whether humans can, in fact, do anything that a computer cannot do. The mathematician Kurt Godel "proved" that math is incomplete, i.e., that there are some mathematical truths the cannot be deduced mathematically. These truths are based on self reference, i.e., statements that talk about themselves. I won't be so arrogant as to attempt to explain that, but there is an interesting book by Nobel laureate Roger Penrose, which goes into it in exquisite detail. The book is: The Emperor's New Mind. In this light, it is interesting to note that Bach's music is recursive in a kind of self-referential way. It repeats itself in different ways at different points in time. Maybe that's an example of music AI can't come up with.

I feel that I'm qualified to contribute to this discussion, as I've been a professional musician for about 50 years now.

We should think about differentiating music generated by A.I. and the performance of it. There is an element of musical performance that is innately human, a personality if you will, that is basic to the way music communicates to us on an emotional level. It may be possible to generate an appealing, even sophisticated piece of music via A.I. but at the present moment I don't think A.I. is developed enough to perform that music in a way that touches us. That has to come from an actual musician. So, in short, yes I would probably listen to a piece generated by A.I. as long as it is performed by a human musician who finds in it something worth expressing. 

When we perceive music, especially the music that we're fond of or interested in, it lights up pretty much every corner of our brains, so our responses are emotional, sensual, and intellectual all at once. Our visual centers, our language centers, our memory circuits are all engaged in this process. Music that is performed artificially (i.e. synthesized) lacks the necessary nuances to accomplish this. We recognize it as "music," but it won't hold our attention for very long.

Perhaps a useful analogy here would be the story of how the test-screenings of the original "Shrek" movie went so catastrophically wrong. The producers had intended this film for young children, but the kids in the test audience got scared and started crying. What happened? Well, they ran afoul of something called the "uncanny valley" effect, sometimes referred to as the 94% factor (and I hope I got that number right). The characters were close to real (94%,let's say), but not close enough. Princess Fiona wasn't seen as a sophisticated cartoon, but instead as a real human with something deeply wrong with her. 

This, I suspect, is how A.I. music would be processed by our very human brains. It might not scare us, but we will subconsciously process it as real music with something deeply wrong with it. 

Have I actually heard any music generated by A.I.? Well, the answer is: I don't know. I certainly don't seek it out, but with music being so omnipresent in our lives I guess I must have encountered it at some point in a movie or TV show soundtrack, or accompanying one or another YouTube video. I just tend to tune that stuff out of my consciousness because it doesn't communicate anything, it's just ambient noise.

The Mechanical

I wonder if AI could make such statements as above and be truly aware of what they implied.