Electronics and computers have a long history in music making, from theremins (patented in the 1920s and most famously appearing in the Star Trek theme song) and synthesizers to the now-ubiquitous electric guitar. Pretty much from the moment we figured out how to produce electrical currents and do work with them in the 18th century, we’ve been harnessing it for play as well. Mostly we humans are actually choosing which notes to play when, although looping, noise, and random composition techniques give away some control over the final sound. But now we’re actually trying to teach computers to compose their own music, with exciting results. You can hear an example as part of this interview about Google’s Magenta project, an exploration of computer creativity across multiple disciplines.
Can computers be creative? I’m inclined to think of creation as an exploratory process, and computers are very good at trying large numbers of possibilities which allows them to explore spaces effectively. That’s how computers got good at checkers and chess, for example; they can try many potential moves in a short amount of time in order to find a good one. Thus, I think computers have the potential to be creative. Where I think the real challenge lies is reproducing the particular aesthetic of, say, music which sounds good to humans. Computers can create all kinds of sequences of sounds and even musical notes, but relatively few of them will sound musical to our ears. Just like toddlers can mash a piano and get notes from it, but rarely does the result actually appeal to us regardless of how encouraging we might be towards the tots.
One of the benefits of teaching a computer how to do a task like composing music is that it forces us to really understand both how we ourselves accomplish that task and how we know when we’ve succeeded. Arguably what has held back computer composition is our limited ability to express concretely and in detail what makes music good. We tend to rely on the Duke Ellington maxim: “If it sounds good, it is good.” Literally, in some cases, as programs would generate music and then human listeners would provide feedback on what sounded musical. What’s different about Google’s new project is that machine learning techniques and computational resources have grown to the point that we don’t have to describe music, we can just give an algorithm a large number of examples of human compositions and the computer can analyze them with enough detail to identify salient features for reproducing our aesthetic preferences.
I’m reminded of the Facebook algorithms that created their own language. Linguistic innovation was not the point of the experiment nor an expected outcome, but it happened nevertheless. And it echoes a point made about language I’ve discussed previously, that language is a cultural artifact which is easier to learn because it can be whatever we want rather than constrained by external factors, so we’ve chosen easy-to-learn languages. I expect music is in a similar unconstrained category. And so I wonder what kind of music computers would sing to each other if they didn’t have to consider human preferences, or human limitations of playing music. After all, Liszt and Rachmaninoff were innovative composers partly because their large hands could play chords other pianists couldn’t manage.
As for the possibility of diminished human exceptionalism should computers demonstrate creativity, I don’t think God is diminished in any way because he created creative humans, so I don’t see why it should be any different for us and computers. And we already see other animals acting creatively, whether they’re making tools to solve problems or inventing games or painting pictures. Granted, some of the more artistic endeavors are prompted by humans and rely on human supplies, and the technical achievement may not be on par with van Gogh or Picasso. Still, I prefer to celebrate the way God’s creation is itself creative.
Relatedly, I’m still looking for another book to read for book club this fall. One possibility is Stuart Kauffman’s reflections on creativity, Humanity in a Creative Universe. I’ve found Kauffman to be an engaging and interesting thinker and so I’m curious what he has to say on this topic. Any thoughts on that option? And of course, if you have additional book suggestions, I’d love to hear those too.
Andy has worn many hats in his life. He knows this is a dreadfully clichéd notion, but since it is also literally true he uses it anyway. Among his current metaphorical hats: husband of one wife, father of two teenagers, reader of science fiction and science fact, enthusiast of contemporary symphonic music, and chief science officer. Previous metaphorical hats include: comp bio postdoc, molecular biology grad student, InterVarsity chapter president (that one came with a literal hat), music store clerk, house painter, and mosquito trapper. Among his more unique literal hats: British bobby, captain’s hats (of varying levels of authenticity) of several specific vessels, a deerstalker from 221B Baker St, and a railroad engineer’s cap. His monthly Science in Review is drawn from his weekly Science Corner posts — Wednesdays, 8am (Eastern) on the Emerging Scholars Network Blog. His book Faith across the Multiverse is available from Hendrickson.