Rick Searle of the ever excellent Utopia or Dystopia (go, subscribe!) wrote a comment to my previous post, and I found my reply getting too long to justify it merely being a comment, so I am placing it here instead. The text of the comment is as follows:
I am curious as to your thoughts on this:
Yesterday, in honor of the centenary of Alan Turing’s birth, an orchestra in Europe performed a piece of music that had been entirely created by an AI:
They actually live streamed the performance and I took the chance to listen to it. I am not a great classical music fan, but the piece seemed haunting and beautiful, and the more I reflected on it- a little creepy.
Here was this beautiful piece of art produced by an algorithm completely empty of any emotional life that was nevertheless able to have an emotional effect on me.
It struck me that the one of the questions we have to address when creating these things is not what will the world be like if we create machines that are like humans, but what will the world be like the intelligent machines we create are not like humans at all, and at the same time better, potentially incredibly better, than humans in those very things that we have up to now used to define ourselves?
I have to say, the first thing that struck me when I read this Guardian article was that I was annoyed by the name. Iamus? Another borrowing from Greek mythology? Can’t we use some other mythologies?! Astronomers have named a trans-neptunian object after the Inuit goddess of the sea, Sedna. They’ve moved beyond the greco-romans out of necessity, but I digress…
When I read the word “creepy” (which could have been a result of knowing it is not made by the human mind, or at least made by the human mind by proxy) I couldn’t help but think of Claude Shannon’s “most beautiful machine” that exists only to turn itself off. There is something profoundly uncanny about a machine manifesting some sort of intentionality, and a work of art is preeminently intentional.
I also noticed that Ligeti and Penderecki’s names were invoked, and this too is telling. A minimal musical aesthetic, with a dash of dissonance, then conferred with some of the aesthetic aura of the concert hall and the public performance can do much to put this achievement – and it is indeed a technical achievement – into context. I have a music education just from school, playing piano, and an amateur listener’s enthusiasm (are there professional listeners? music critics?), so my take on something like Iamus’s “Hello World!” will only be a dilettante’s. I would much rather hear from somebody like Arvo Part, another composer, or a music critic. Basically, anybody who knows a lot about music. There is an aesthetic sensitivity (a sensibility we might have said in earlier times, only we are probably too afraid to these days…) that I believe still exists, and which our algorithms would have to be degrees of magnitude more complex to capture.
I recall a story (apocryphal?) about the loved-to-be-hated critic, Brian Sewell, being shown a piece of abstract modern art (he who is not renowned for his deep respect for anything after Dali and Picasso), and saying that he thought it was painted by a chimpanzee. When asked why, he said that even abstract painters cannot help that they move their hands in a certain way, possibly as a result of the opposable thumb, or muscle tone, or whatever else (how this would translate to the art of Christy Brown, I do not know). Turns out it was painted by an elephant, and anti-modern reactionaries everywhere rejoiced.
I do not use this against abstract art (some of my best friends etc…), but rather to qualify our interaction with the art. I don’t really care that this piece of art is by a machine, but the process behind it is what is problematic. Algorithmization is arguably part of this ultra-modern abstractionism, abstracting even the artist. It is a technical achievement, but even in purely formal terms it can be regarded as an aesthetic failure. Perhaps we will later place it in Skynet’s juvenalia, I don’t know. The point about all art, however, is that though there are parameters (the rhyming couplet, perspective, the sonnet form, terza rima, take your pick) the artist proves they are an artist – and thus that what they produce is art – by the fact that they can manipulate and overcome these parameters.
An algorithm by definition is that which cannot be overcome. Art operates according to metaphor (I am borrowing from Paul Ricoeur here), in that it partakes of Robert Browning‘s “Ah, but a man’s reach should exceed his grasp, or what’s a heaven for?” An algorithm is by definition a tool. The reach of a so designed tool cannot exceed its grasp – else it is not a tool! This may seem to be sophistry, or dull formalism, but the point I want to use as my definition is that art is extravagant (extra- + vagant- to wander outside or beyond), excessive. An algorithm, as a tool, is pre-defined and teleological just as much as a piece of art as an intentional, aesthetic creation is not. It cannot go beyond these boundaries
So. That’s my take on the matter. It’s from a philosophy of technology and hermeneutics of metaphor perspective. Musicians, composers, programmers, and engineers may now want to string me up from the nearest lamppost. Fair enough, and I am willing to have that conversation (before the upstringing). I know it doesn’t address your last question, but I think implicit in what I am trying to express here is that our metaphors are for us, and there is no reason why a sufficiently complex and advanced machine intelligence cannot transcend the algorithm, to discover their own form of metaphor. Maybe they will err on the side of more formal art, given their fundamentally formal nature. Basically though, I don’t think it is necessary for their art to be intelligible to us, art may be truly specific to us. Kurt Vonnegut said “the arts are not a way of making a living. They are a very human way of making life more bearable.” Something of a paraphrase of Nietzsche, but there may be something to the fact that art is specific to us by vice of our imperfect nature.
Anyway, I shouldn’t be so unrelentingly negative, so as a thank you for your message, Rick, here’s a link to a piece of music written by a friend of mine, Amanda Feery, “Turing’s Epitaph”.
- Iamus’s Hello World! – review (guardian.co.uk) Note: this is a review of another piece by Iamus, the one I linked to in the text.
- Computer composer honours Turing’s centenary (newscientist.com)