A.I. Isn’t Going to Replace Your Favorite Band
7 min read

A.I. Isn’t Going to Replace Your Favorite Band

Hello, if you’re a new Patreon subscriber and want stickers from me, please send me your address so that I can mail those out by the end of the week! This is slightly outside the realm of music but if you or anyone you know does illustration work, check out the new website Litebox, which crowdsources rates for jobs and is looking to do cool work within the illustration space. Lastly, I was sick last week and tried to get a bit of rest, so this newsletter on artificial intelligence is a little shorter than I initially intended. Just had to be honest!

Earlier this year, Holly Herndon, one of music’s most critical thinkers, put out her latest album PROTO, which received quite a bit of press, not only due to its quality (it’s great!) but also because she worked with an artificial intelligence program (which she named Spawn) on a number of tracks. What I appreciated was that Herndon’s discussion of the technology is that she actively leaned against the marketing framing of artificial intelligence as a means to replace human labor and instead contextualized it as a mere extension of it. Herndon speaking to Jezebel said (emphasis mine):

That’s one of the biggest problems of AI; it’s this kind of opaque, black box technology, and when we have this glossy press release where it’s like “the machine just wrote this song” you’re totally discounting all the human labor that went into the training set that the thing learns on. That was a really important part of how we set up the project and the way that we did. We wanted the people training Spawn to be visible, to be audible, to be named, to be compensated because I think that’s a huge part of what we’re facing with this thing today.

I found this particularly interesting because new advances in technology tend to be framed as simply a replacement for the human labor force. Outside of music, this leads to a parade of headlines implying that robots will replace human workers and that there is nothing that one can possibly do to stop this. Even within music, this is an old trope as back in the 1930s, during the early days of recorded music, the American Federation of Musicians ran a massive advertising campaign warning against the risk of poorly-recorded music coming to replace good-quality live music. Capitalist hindsight would claim that the union was foolish to fight against such technological change given the outcome of history, but I’ll argue later there's a perfectly good reason, as a worker and as a society, to more deeply interrogate who truly benefits from such technologies.

In October, on the New Models podcast, Tom Krell of the How To Dress Well project made an aside about how Spotify’s end goal is simply to get rid of musicians and create entirely A.I.-generated music. The nihilistic future he tosses out is generally agreed upon as if the entire music industry is on a death march towards automating away a basic mode of human expression. That discourse is so dystopian that it’s what inspired me to want to push back.

(I’ll just make a quick aside here — one of my favorite podcasts, Citations Needed, released an episode last week around the media’s narrative of automation, which I’d highly recommend since I borrow spiritually from a few of its conclusions.)

Frida Garza, in Gizmodo, cited a late 18th century Wolfgang Mozart “musical dice game” in which numerically-generated pieces of music are strung together for one to perform, like an early version of what would become Krell’s nightmarish muzak vision. Garza more specifically tracked these computer-based music dreams back to a couple of professors at the University of Illinois at Urbana-Champaign in 1957, who experimented with the idea early on. The technology continued to advance over the following sixty years but over the past few years,  a couple of flashpoints suddenly appeared at the intersection of artificial intelligence and music.

In 2016, Sony’s Computer Science Laboratories worked with songwriter Benoît Carré to produce a song in the style of the Beatles called “Daddy’s Car”, to which the Verge gave the obviously glib headline: “This AI-written pop song is almost certainly a dire warning for humanity”. A year later, François Pachet, the project’s director, left for Spotify. He helped lead the project SKYGGE, which collaborated with the singer Kiesza and got slotted onto a number of Spotify playlists, in order to promote the company’s pet project. If Krell’s fear is that Spotify will replace musician labor with robots, then that’s already happened. However existential that concern may be, it was again triggered when Spotify was caught putting muzak onto popular playlists. That I find worth exploring a bit more.

Do Androids Not Dream of Historical Facts

Earlier this year, Cherie Hu interviewed Alex Mitchell, the founder of Boomy, a company that allows people to quickly produce music and monetize it through traditional for-profit music streaming platforms. There was one particular quote from Mitchell that stuck out to me for its absurd ahistoricism:

But if you just look at the history of music and technology in general, there’s never been a significant advancement in music technology that was not immediately met with fear, and eventually was considered something that is normal, and a thing that everyone does. Right?

Just to throw some history out there, the early introduction of player pianos and, eventually, records shifted home music from being an active engagement into one of passive consumerism. This isn’t to glamorize the 19th-century period where, of course, only families of a certain class could read music and afford instruments and learn to read music but the introduction of recordings did interject a product into a space that previously existed for human expression, which is often under-considered with the sheer ubiquity of recorded music. A similar shift occurred when movies introduced recordings; this eliminated the jobs of over 20,000 unionized musicians who worked in the pits of movie theaters across the country. Earlier this year, I wrote about Damon Krukowski’s Ways of Hearing, where he highlights the volume of tangible audio details that were lost in the shift from analog to digital audio. In contrast, Mitchell’s use of the word “normal” hand-waves over the real cultural losses and trade-offs that took place with the introduction of new technology that sought to increase bottom lines, not foster a deeper appreciation of music.

This is why I actually wanted to ground this brief newsletter with that opening quote by Holly Herndon on the implication of labor within this technology. There is the commercial side of artificial intelligence that will lead to endless pablum by start-up founders, blathering by industry commentators without a clear vision of how A.I. truly may deeply change the industry. Or there can be cool stories like one the Verge reported about an open-source vocal isolation program that was started at Deezer, but is now available to anyone but to bring back that issue of human labor. The Verge mentioned in the article (emphasis mine):

This tool seems extremely capable but be warned: you’ll need some tech expertise to use it. Unless you’re regularly playing with software like Python or Google’s AI toolkit TensorFlow (which was used to train Spleeter) you’ll have to to download a few programs to get Spleeter up and running. And you’ll have to comfortable using a command line input (albeit a very simple one) instead of a more accessible visual interface.

This isn’t to discourage anyone from using it but rather to show that this new technology still arrives with a new set of barriers. Lastly, there was the on-the-ball question that Hu posed to Mitchell in her interview about the implications of this technology when its primary players are Google, Spotify, and other large tech firms. More concerning than how song accreditation would work or what to do with a flood of A.I. tunes is a rather consistent theme in this newsletter where I raise a red flag to ask: What happens when an entire medium is consolidated into a few companies? Again, concerns about A.I. taking over human musicians aren’t rooted in paranoia over technology but rather in accessing the capitalist logic of cutting costs even if it eliminates the entire creative process.

Unheard Labor

In case you’re wondering how many Spotify streams would be needed to cover one’s costs of living, check out this graph. Kickstarter’s union still hasn’t been recognized and former Kickstarter worker Taylor Moore spoke to Jacobin about his experience organizing there before being fired. I also saw a nice article on Mic about #NoMusicForICE. Lastly, I’d like to note that the American Federation of Musicians is rallying in the Chelsea neighborhood of New York City tomorrow (November 7th at 12 pm) to demand fair streaming royalty rates.

Sofar, So Bad - The Baffler

Liz Pelly on the live music grift known as Sofar Sounds is a must-read. I absolutely cannot recommend it enough.

Streaming Growth Is Slowing at the World’s Biggest Music Company. Could TikTok Help? - Rolling Stone

My gut says that TikTok is slightly overrated. Yet, when I read these stories where TikTok is framed as being the next cash cow for the entire music industry, it reeks of desperation. Major labels are just seeking out (unsustainable) growth as Spotify starts to dry up.

Bytedance faces US national-security review of Musical.ly acquisition - Music Ally

Relating to my previous point, even if I hold skepticism towards TikTok’s long-term business viability, I am all in for following the endless “Red Scare” paranoia around the company and the United States government.

Is Bollywood losing its dominance of the Indian music industry? - Music Ally

If you ignore the headline, this is actually a great piece that discusses the last few decades of Bollywood music and the emergence of non-Bollywood Indian music in the mainstream. What I find interesting is just how many companies are pushing non-Bollywood tracks, as if to show that while there might be a slight change in aesthetic, the industry power-players remain the same.

Which is the best streaming service for supporting artists? - The Guardian

The reason I’m so obsessed with the user-centric model for streaming is that there is an existential question (the title of the piece) that music fans want to have answered in a satisfactory way. Again, I’m not saying that the user-centric model is the final answer, but it's certainly better than what’s available at the moment.

Global Anti-Piracy Coalition Battles an Overlooked Enemy ⁠— Password Sharing - Digital Music News

I’ll continue to sound like a broken record by re-stating that piracy is truly just a scare tactic used by wealthy companies to blame and shame consumers who likely cannot afford expensive media subscriptions but still consume their favorite music or television shows in other ways. So, the idea of cracking down on password sharing is certainly absurd but by no means any more absurd than previous anti-piracy efforts.