What does AI mean for composers?
Well, it’s here. AI capable of generating music that almost sounds good is here. It’s still unreliable, but it’s improving rapidly, so music makers should be thinking about the implications. On one hand, we will be able to harness AI in our work to explore new musical possibilities. But there will be downsides. What types of work will be most vulnerable? What types of work will flourish? What can we do to help each other? And might there be a silver lining?
The first type of work handed to our AI comrades (and for the record, I have always been a huge supporter of our robot overlords) will surely be music for commercial settings. Background music for corporate videos, advertising, YouTube videos. Corporations don’t care about supporting composers when they make a compulsory training video for new employees. Finance YouTuber X doesn’t need a human touch in the underscore of their video.
With tight budgets on all but the biggest films and TV shows, it is easy to imagine simple soundtracks being farmed out to AI too. Some of the best music ever written was written for film. But, by its very nature, film music draws on tropes to communicate: pizzicato strings are playful and low drones are suspenseful. This basis in communally understood sonic signifiers means AI could potentially produce reasonable film cues if given clear prompts.
The above highlights an uncomfortable reality. Some of the first to ditch creatives in favour of generative AI…may well be other creatives on a budget. Composers wanting an image for the front of their score, film makers wanting a soundtrack, visual artists needing sound design for their installation. I am already seeing creatives turning to AI in search of ‘supporting art’ to decorate their centrepiece creation. It seems a bit perverse to me and I really think we should be looking out for each other in this regard.
But what about music for the concert hall? The kind of music I am personally most involved in. Living composers of concert music are already competing for program space with three-hundred-years’ worth of excellent dead composers. We don’t need any more competition! Please, for the love of Terry Riley’s beard don’t let there be hordes of brilliant robot composers who can remember every oboe multiphonic. I’ll be ruined.
Perhaps I am in denial, but the impact on concert music feels less immediate for two reasons:
- First, the level of detail in contemporary classical scores is often high. That doesn’t make it better or worse than any other music, but may make it more difficult to mimic. Concert music is also designed to have an audience’s full attention in a way that background music for corporate videos is not. AI concert music will have to be pretty polished before it’s fit for purpose.
- Second, I don’t sense an appetite for AI generated music from ensembles and audiences. This one’s a biggie. I just don’t see any market forces pushing classical ensembles towards programming AI compositions. Sure, if an orchestra commissions a computer instead of me they might save $100,000 (yes that’s my normal fee, why do you ask?) but who will be socially awkward at their post-show function? I’m much better at that.
However, concert music composers won’t be able to ignore AI. In fact, I think our work will increasingly exist in direct dialogue with it. As AI becomes better at replicating established forms and genres, we’ll be under escalating pressure to offer music than ‘couldn’t have been written by a computer’, music that is somehow personal and idiosyncratic. It could super charge the rate at which the composing community innovates.
More than anything, the possibility of AI generated music at the click of a button will surely lead composers, ensembles, and audiences to place greater value on the process of creation. When we present new commissions, we will surely talk more about the way the music was created, the way the collaboration between composer and musician functioned. There will be an imperative for new pieces to be unique to the time and place they were made, and the community that made them. This might even be a silver lining.
