Yesterday, I was fortunate enough to join Joe Henson, Alexis Smith, Lydia Gregory, Martyn Ware, and Henrik Oppermann, along with our host, Tony Nwachukwu from CDR Projects, on a panel at We Are Robots 2017. Our discussion was on Music and Interactive Entertainment, which quickly spiralled into an emotional debate about the relative merits and extreme negatives of the use of technology, particularly artificial intelligence and machine learning, in music.
This was, it’s true, to be expected. The general consensus within the music making community is that such levels of technological intervention interfering in the very human art of music is a travesty. I happen to agree. I am far from a musician myself, but I am a listener. And I have long held the belief that the increased use of technology in sound production has been a double-edged sword, and one which – from my own perspective – offers me little pleasure.
I have never enjoyed electronic music, on the whole. That’s not strictly true when it comes to certain genres, particularly hip hop, in which sampling is rife and actually adds a great deal to some tracks. However, my heart has always been in the realms of bands like Led Zeppelin, Jefferson Airplane, The Rolling Stones, The Beatles, The Small Faces, The Velvet Underground, T Rex, Jimi Hendrix, and then Fleetwood Mac and Bob Dylan (on another note). My feeling has always been that the further we have gone down the rabbit hole into the digitisation of music, the more we have lost the raw emotion of it. I summed this up to some extent on one comment I made during the panel discussion:
— Tiago TC (@tiagotcorreia) November 2, 2017
I am more than aware that many would ardently disagree with this sentiment, and I accept that. As I’ve gotten older, my view has softened somewhat, and it has had to, particularly in light of my growing love of hip hop.
That being said, for me the love of music has depended principally on the use of guitar (preferably electric) and, above all, vocals. Perhaps this is something to do with being a writer, and having a fondness for poetry which was actually born out of my adoration of the lyrics of my favourite old musicians. It’s this same fondness that has drawn me into true hip hop as time’s gone on.
As I say, digital music largely leaves me cold, so whilst I sat on that panel playing devil’s advocate for AI, I had one big question: if musicians are against using AI in music, then where do they draw the line? Most musicians use tech to some degree in their creative process, as did my own old favourites whose music, although raw and viscerally human, was not produced without some manipulation, even if that be simply a wah-wah pedal. In fact, hang on, the electric guitar itself is electronic. So when I say I don’t like electronic music, where do I draw the line?
Drawing The Line
It seems that the fear of eradicating human input is at the heart of the aversion to AI. This is the same as it is in pretty much every industry, but it is the creatives that feel the pain of that most of all. Our human-ness means so much to us – we are, as a group, generally more connected to the notion of the soul and to emotion than most people. Or, at least, we like to think we are! If we allow artificial intelligence into the fold, then we are no longer special. Maybe that’s just our egos talking. As Lydia Gregory said during the panel,
“It would be sad if the thing that stops us from adopting music technology is our own vanity”.
A lot of the ire that reared its head during the panel discussion focused on the use of machine learning in generating music for media purposes (such as to accompany adverts, YouTube videos, movie scores, etc). Where does this put the musicians for whom the creation of stock music and soundtracks is their livelihood? It’s hard enough to make a living as a musician, and as my fellow panellists noted, takes a long time to build a career in. It’s seen that those who make money from their art are the lucky ones. The rest of us have to try and fit our craft in around our day jobs.
Don’t get me started… pic.twitter.com/aUrb3phnIy
— The Fright (@theflightmusic) November 4, 2017
I know, from my own experience, that the longer I sell my writing as a commodity, the further I get from my roots. The last poem I wrote was to read at my best friend’s wedding, and it was agony feeling as though I’d lost my touch. If something cannot be monetised, it’s seen as quite useless, pushed to the sidelines in our society. It’s inarguably sad, and the idea that human creativity could be pushed even further from our lives in favour of machines is a terrifying thought.
There is no objective measure of how ‘good’ a piece of music is. Functional music, for certain low-grade soundtracks, ‘hold’ music – general ‘muzak’ – is probably the best use of entirely AI-generated tunes of the sort composed by platforms such as Jukedeck and AI Music. Nonetheless, it still puts the human creators whose income depends on creating this music in career jeopardy.
But the fear that AI music will actually replace human creativity is bullshit. Will Darbyshire argued the case very eloquently in this piece about the difference between knowledge and understanding.
Earlier in the year, I interviewed artist Mat Collishaw, on his exhibition Thresholds, in which he used virtual reality to recreate the first photography exhibition from the 1830s. I asked him his thoughts about the use of technology, particularly VR, in art. This is what he had to say:
“I am not particularly enamoured with technology in general, not least with the notion of virtual reality casting its shadow over the art world. That being said, I am not sure that VR can actually compete with the lived experience of art. The gallery experience is meditative, almost church-like. You wander and engage thoughtfully with the works around you. There’s a sense of physical connection in the gallery or exhibition space. Can this subjective experience really be translated to a simulation? I don’t think so. Whilst the threat of automation in industry is quite real, I think the art world will survive the ravages of technology.”
Fine art is not exactly a democratic medium in the way that music is. Music is usually made for listeners from all backgrounds to enjoy, whilst understanding and enjoyment of contemporary fine art is often limited to those with the educational means to unpack what they’re seeing. Though on a purely visual level, anyone can enjoy fine art, the truth is that a lot of people go away from their gallery experience confused and often complaining that ‘modern art is rubbish’.
Music in popular culture is profoundly different to that. The wealth of genres and moods of music to be consumed can appeal to anyone, and therefore most – if not all – people have an opinion on individual songs and artists across the board. As music is, in its essence, an accessible medium (as a generalisation: of course there is high brow weird stuff that confuses all but the most elite musical intelligentsia) there is also the pervasive and long-standing complaint from the musical community about the ways which big labels strangle off upcoming talent and suffocate the emergence of new movements.
The music industry is big business. Less so now than it was, since the advent of the mp3 and then streaming services, but still there is big money being made. But, as everyone knows, these big labels have certain criteria for what sort of bands and musicians will be granted a contract. They must be marketable. Many (x) factors have combined to create this sorry state, and the use of technology in the form of big data, algorithms in streaming services, digitised music files, etc etc, has certainly played its part. Henrik Oppermann made a great point about this during the panel discussion:
Spotify’s Discover Weekly generates playlists based on listeners’ individual music tastes, which is a double-edged sword. This algorithm has the potential to thrust lesser-known music into the ears of those who would never have otherwise come across it. But on the other hand, it is more likely that it contributes to the blandification of music, exacerbating the ‘echo chamber’ problem that such algorithms are responsible for across social networks and search engines. The big data approach may be useful, but it is blind to the subtleties of the job it’s ostensibly designed to carry out.
MySpace started a revolution in that before being snuffed out by the big bad technocracy of Facebook. Every musician could have a page with their tunes available to listen to on MySpace, and many well-known artists were actually discovered there. We have Soundcloud and Mixcloud now, as well as YouTube, so the social network side of gaining exposure for musicians is great. The same issue of labels only snapping up the talent that fits their formula, however, remains. Although platforms like these give more exposure to small-name artists, there is no money to be made from listeners alone.
Musicians, like everyone else, need money if they are to continue producing their art. And money is a hard thing to come by. Most musicians make their money in general tedious day jobs and spend their nights playing small venues and open mic nights, recording for the love of it, hanging out with music buddies. The act of making music, something that musicians would like their entire lives to be about, has to be pushed to dusk and conducted in the shadows. If it weren’t for the need to make money to support themselves, music could be a much bigger thing altogether! The cost of living is higher than it has ever been, leaving still less time for passion and emotion in life. Things are grey in this technological era where we are promised colour and freedom. There’s a fundamental disjunct between what we’re told our lives are and will be like, and what they really are.
I’m Alright Jack, Keep Your Hands Off Of My Stack
Maybe the digitisation of music files has a lot to answer for in reducing the money to be made for artists. Whilst those of a certain age remember eagerly running to the local record store for the latest release, whether that was on vinyl, cassette, or CD, younger generations do not know this pleasure. The joy of holding a material item containing the sounds you love, accompanied by beautiful visual artwork on the cover, is something I lament my own child not having the opportunity to experience. Some of us hold on to our record collections like babies, and continue to grow them with the same fervour as always, and independent record shops still scrape by, but the mp3 and streaming services have all but killed that little bird of paradise.
Spotify is famous for paying a pittance to artists on its platform, as millionaire pop singer Taylor Swift made a big noise about before she relented and moved over to the dark side. Even The Beatles only got their music up there last year. I’ve lost access to Bonzo Goes To Bitburg by The Ramones from my Spotify recently, which I assume is part of the same row. But these are all big names, big names who have a huge amount of money already, so their complaints about not making money from the plays of their tunes raises my eyebrows well into my hairline.
It’s the little guys that need Spotify to get heard and recognised. But they can’t or won’t go there because there’s no money in it for them.
Spotify itself is little stranger to cash flow issues. The company has still yet to report a profit, as year after year its net losses grow. The platform continues to have problems with monetising its offering. Popular as it may be with users, it’s not enough to grow the company. At some point, it’ll go bust, unless a bigger company with bigger ideas are able to crack the monetisation code and buy it up. My bet is on Facebook.
Money is one of the key points in the whole music tech dilemma. It’s not that there is less money for people, per se, it’s just that that money is being funnelled into other areas. Martyn Ware made an interesting point about the parallel industry, film. The creation of CGI versions of famous actors, for example, is something that’s becoming all the rage. Using the technology because it’s there, not because it’s necessary. Speaking to Tiago Correia after the panel, he expanded on this point:
“This is the bottom line as to why most AI is being pushed: automation brings down costs.”
Rather than treating technology in music as a kind of us versus them battle, maybe musicians have an opportunity to evolve through collaboration. Whilst there are some technologies out there, namely music generated by machine learning algorithms and sold wholesale, that leave a lasting bitter taste, there are others that are actually helpful.
New tools, such as Voclea, which is a machine learning algorithm that helps artists to articulate their creative ideas and capture those ideas as they happen. By making sounds with your mouth, the algorithm will translate them into an instrument, back them up with a drum beat and melody, thus suggesting ways in which your idea could be expressed. There is, of course, no obligation to use what the algorithm comes up with, so it’s essentially not much different than playing a riff you’re thinking of to a friend and them coming back with “how about this…?” Another ML tool by PopGun uses an AI named Alice, who will play a duet with you – you play some notes, Alice plays back with what it thinks will sound good next.
These are just two examples: there are many more. There is much to be said for artists seeing past the obstacles that some uses of machine learning in music are causing, and focusing on collaborating with tech that can help develop new sounds and ideas. After all, that’s what’s been going on in music for decades now. As Lydia Gregory noted during the panel discussion,
“I doubt Bjork would be making the innovative music she is now in the 19th century”.
We talked about Bjork quite a bit on the panel. Whilst I used the artist’s work as an example of widely lauded music that uses a lot of tech in its production to ask where we draw the line, my fellow panellists identified the key point about Bjork. You never know what she is going to do next: she is evolution of music in action. Every new thing she does brings with it a thoroughly new experience for listeners. Her work is the opposite of more mainstream artists. As Taigo Correia remarked:
“No one knows what will come out of [Bjork], but I don’t think it’s because she has ‘soul’. It’s because she’s Bjork, [whilst] I’m pretty sure I know what’s coming out of the next Katy Perry album, for example.”
It was Brian Eno who famously popularised the term ‘generative music’ to refer to music that is “ever-different and changing, and that is created by a system”. But perhaps, as one of my fellow panellists noted, the world doesn’t need any more generative music now. After all, “Eno has created it all”.
A lot of the emotion fuelling the panel discussion was born out of this mindset of “I just don’t understand why anyone would want robots to do music”. As Correia said to me afterwards, this is not necessarily the most interesting position on the debate. As with anything, there is no point seeing the issue of technology in music, particularly things like AI, as a black and white matter. It’s a sliding scale of positives and negatives, albeit those that artists will have to work through in order to find a workable solution.
There is a very real concern that such technology will allow business to further control music production and continue to monetise and monetise the art, but this will always be only one facet – you cannot kill the human art. However, you can evolve it, by giving in to your resistance and using the new tools available to your advantage.
My thanks to Tiago Correia for his input for this article, and for reminding me of some of the key points discussed that my memory failed to recall. My additional thanks to Gordana Jovkovic, Emma Joyce, and all the team at We Are Robots for a fantastic event, and to my fellow panellists for a fascinating and lively discussion.