Tuesday, October 27, 2009

Robophobia in The Grauniad

Last week's Guardian has a blog piece by Mark Lawson about the composer Emily Howell, who has a forthcoming album. Part of what is interesting about Emily is that she isn't a natural person, but a trained computer program. She's the successor to Experiments in Musical Creativity (EMI, or 'Emmy'). Both are projects of David Cope, himself a composer. There's a lot you can read (including a lot that’s on line) about the history of Cope's work, and the music produced by the systems he has developed and trained. I'm not going over any of that here, but I do want to take issue with some things Mark Lawson has to say. Lawson doesn't care how good Emily's music might sound. He says it's "worthless" because it's not made by real people. I think that he is being a silly anthropophile robophobe twerp.

After some rather disorganised paragraphs in which Lawson half-heartedly faces up to the fact that much creativity is a matter of re-arranging elements that are not themselves original, he turns to banging the table. Along the way are a few telling bits of rhetoric that show honesty is not a big priority for him, including the gem that when Emily produces a work it is by “simply randomly reshuffl[ing]” bits of another. Clearly he’s simply ignored the fact that Emily is laboriously trained, and that the process of construction is guided by the set of constraints produced by the training. Anyway, here comes the table banging:
So logic is on her side. Art, though, is illogical. Although she can be defended intellectually, the creator of From Darkness, Light is no more a composer than a synthetic sperm knocked up in a laboratory would be a father.
Oh. So a traffic light isn't a “real” instruction, because it's just a machine. Calculators don't tell us arithmetic truths, because they're not people. I haven't really been to Scotland because I didn't walk there. It’s not about what happens, it’s about where it comes from. Why should we think this? Lawson continues:
Music, writing or art is a communication between two humans. This does not mean it has to be emotional or warm – a delusion industrialised in large parts of Hollywood – but that there is some sort of conversation between two members of the same species, even if the artist's side of the exchange is "go away and leave me alone".

Paradoxically, it was JD Salinger, a novelist who has refused any rapport with his readership outside the pages of the books, who most beautifully captured this truth when the narrator of The Catcher in the Rye suggests that reading a really good book makes you want to phone up the author. A composition by Emily Howell might make us want to email her, but we know that she could not reply. Admittedly, we also know that Salinger wouldn't take our phone call, but the crucial difference is that he could if he wanted to.

A computer, cleverly programmed, could probably produce the Doubting Thomas Passion by JS Bach or More Snow on Kilimanjaro by Ernest Hemingway. But the exercise would be worthless because the works from the software would not be informed by being a God-fearing kapelmeister in 18th-century Germany or a suicidal macho male in mid-20th century America.

Our shelves may be full of composers and writers who could be accused of having only artificial intelligence, but their efforts are still more worthwhile than art created by AI. "From the heart – may it go to the heart," wrote Beethoven on the manuscript of his Missa Solemnis. From the byte to the brain can never be equivalent to that.
This is a pretty strong set of claims. No matter how beautiful, or moving, or exciting, or anything else music, including Emily Howell’s, might make us feel, it's "worthless" because it didn't get to us by being passed through the brain of a natural person. And the reason for that being important is that we could (in some fabulously diluted sense of could, which covers long dead people who can't talk to anyone now, and living ones who don't want to talk to us, etc.) talk to them about stuff.

It's interesting that no matter how it got here, and irrespective of whether anyone could (or would want to) talk to where it came from, that is surely the most lousy justification for a claim that I've seen all week.

Lawson is, I’d argue, perfectly free himself to have a silly prejudice to the effect that he prefers music that in some sense came out of a brain. But it’s arrogant and absurd for him to declaim that such much is in general “worthless” just because he’s prejudiced.

More than that, it’s confused. People, and their brains, are physical systems. Their basic working parts are mechanisms – mechanisms of DNA transcription, protein construction, ion channelling, neurotransmitter action. Their interesting functions are the product of gigantic co-ordinated action among these myriad mechanisms. This means that if having in some sense been produced by mechanisms guarantees being “worthless” then everything made by any person is worthless.

Besides all that, it’s fascinating to learn more about music, and what sorts of process can compose it. There’s little reason to think that what goes on in Emily will be strictly analogous to what happened in Bach’s skull, but there’s at least a tantalising suggestion that we have more idea that we used to about what might have been in there. And there’s exciting work to be done – I for one would like to see computers capable of sophisticated ensemble improvising.

Here are some links:
Lawson’s article (comments unfortunately closed).

David Cope's mp3 page, with material by EMI. (In particular, see 5000 works in Bach Style.)

Article on Ars Technica.

Article on Times Online (includes streaming media with short clips from the forthcoming album).

Article on Vox.
(The image at the top was lifted from an image challenge on B3TA.)

No comments: