Pop music: Is there a threat of an AI apocalypse in music?

With the Dall-E image generator you can create images that can hardly be distinguished from man-made originals. And it can hardly be denied that ChatGPT can already keep up with one or two averagely talented novelists. And the AI ​​is already knocking on the next door with music generators like Sonu AI. In a matter of seconds, the new programs can now be used to create songs in any genre and with lyrics of your choice that would hardly be noticeable if they were played on the radio. No ability, no creative thinking ability that was previously reserved for humans no longer seems safe. Art is obviously no exception.

In the music industry, this has now apparently also become clear to artists. In an open letter published last week, over 200 US musicians, including stars and prominent bands such as Billie Eilish, Katy Perry, Stevie Wonder, Peter Frampton, Smokey Robinson, Pearl Jam and REM, warned of the dangers of AI that represents an “existential threat” to their art. This is remarkable because so far many of the signatories have been able to live very well with a exploitation system in which many artists’ existence is already threatened.

The fact that even the big earners in the music industry are now warning about the new technology shows how seriously this development should be taken. Ultimately, the use of AI programs in art does not only affect artists. Art as a (re)producer of culture and social reflective practice affects us all. It’s tricky: Technology should free us from mindless Excel spreadsheet and assembly line work and give us time for the beautiful things in life. Instead, we humans just continue to work as usual while DALL·E paints pictures, Sonu composes songs and ChatGPT writes novels.

But perhaps it would be helpful to counteract the knee-jerk discomfort with a more differentiated view? Otherwise, prophecies of doom become a self-fulfilling prophecy that we humans only follow in self-imposed helplessness. Using the example of the next big thing in AI art, music generators, we can determine whether the future really looks as bleak as many fear. Because what is the concern? That the radio and the big streaming playlists could only play a uniform mash of soulless, templated songs? This has long been the case in many parts.

In his essay “Capitalist Realism,” the cultural scientist Mark Fisher described how, in contemporary capitalism, market mechanisms ensure that commercial music becomes increasingly similar and the music industry repeats financially successful concepts over and over again in a formulaic manner until at some point it hardly produces anything new anymore. A thesis that is currently confirmed by a study for which 12,000 English-language songs of various genres were examined, with the result that the song lyrics have become increasingly simple and similar from 1980 to 2020. Ultimately, in a market focused on efficiency and profit, it always makes more sense to repeat a concept that has already been successful rather than investing in projects with uncertain profitability.

Artificial intelligence could even speed up this process. This is due to the way it works: based on huge data sets, it calculates how likely one element follows the other and generates a result from the corresponding combination. In language models like ChatGPT, the AI ​​generates sentences by repeatedly putting the most likely word after each word. Models whose data base consists of music use the same mechanism to place one note after the other until an entire song is created. In a future where music generated entirely by AI is marketed, this would mean creating new music from the similarities of existing music. To merge the same with the same in order to obtain the ultimate same. This is a very attractive scenario for the music industry, because the changing faces that have previously marketed their music and served as a projection screen for consumers have so far had to pay labels not only for their appearance, but also for their music. AI could change that in the future.

What does this mean for musicians themselves? You have to know that AI-fication was preceded by another development in the last decade, namely the digitalization of the music world and the associated enormous differentiation of musical styles and genres. While things became more and more uniform at the top, the base became more and more diverse. Due to lower costs and digital distribution options that are available to almost everyone, the barriers to entry for musicians have become lower. At the same time, the major streaming providers, who have made this development possible, pay according to the Matthew principle: whoever has, will be given. Anyone who doesn’t belong to the top percent of the music business can hardly make a living from music alone these days. Spotify pays just under 3,400 euros for a hit that has been listened to a million times.

The higher up an artist is in this distribution mechanism, the greater the pressure to adapt. If a label invests a large sum in an artist, the risk should remain as minimal as possible. There is no time or money for experiments. But what will happen if the market is flooded with generic, AI-produced songs? Their value drops. On the other hand, money that labels previously invested in elaborate productions with expensive artists is freed up. Labels would only have to market familiar faces and turn unknown faces into familiar ones, while the AI ​​delivers the right music. Although the current top earners in the streaming business will probably not have to worry about their existence, this look into the future explains why musicians like Nicki Minaj and Katy Perry are now also thinking about distributive justice in the music industry make. Your negotiating position doesn’t get any better with an AI breathing down your neck that can produce almost identical songs in a matter of seconds for virtually nothing.

This could have the effect that inflationary AI music produced according to established formulas increases the pressure to innovate and the value of music produced outside of the mainstream formulas. Anyone who has ever listened to newcomer lists knows that uniformity is not just a phenomenon of the charts. And an aspiring artist who can create a radio-ready song on the Internet in a few seconds might think twice about whether it’s worth reproducing the exact same thing herself or trying something new. With a massive oversupply of uniform songs, musicians far from the beaten path could regain importance. Human-made could become a unique selling point – and the underground could become the organic farming of the music industry.

The AI ​​apocalypse is not a sure-fire success. The grievances lie just as much in the present as in a possible future. In a way, AI is accelerating a distribution struggle in which the majority of artists are already left empty-handed. Artificial intelligence only becomes important in the social structures in which we embed it. It does not exist apart from it. Technology in itself is never harmful or good, but always what we make of it. In order to give art and culture and those who create it a worthy place in our society, we may not have to wait for the great AI apocalypse.