Competition with the (Music)Machine (1)

Competition with the (Music)Machine (1)


The question that arises when it comes to AI and human creativity (which will only be answered exactly in the coming years), is how the value of human work will assert itself in competition with algorithms.

At the moment, many fear that they will lose their jobs and be replaced by a „machine“. But it’s not that clear – we don’t lose our jobs automatically, but only if we lose out in direct competition with the machine. In our capitalist society there are strict laws that are shaped by supply and demand and the success of the most efficient production method with maximum profit. But there are different gradations – a mass product must be produced as cheaply as possible so that it can be bought as cheaply as possible by as many people as possible. A luxury or niche product, on the other hand, benefits from its rarity – it is produced with a little more effort, but can also be sold for significantly more money.

A craft beer manufacturer, for example, who brews her beers with great attention to detail in small margins according to her own recipes and sells them locally will never compete with a mass producer. But she may still be able to pay her few employees very well and live from her small produce. Perhaps the employees of the small brewery even earn much better than an assembly line worker at a large brewery. The head of the craft brewery, on the other hand, will earn only slightly more and not even close to as much as the CEOs of the large brewery.

However, the craft beer manufacturer also needs the large brewery. Because if there weren’t a major brewery with cheap and therefore poor-quality beer, the craft brewery wouldn’t be able to set itself apart and would have no reason to charge more for its beer. So there is a mutual dependency between average, above-average and, of course, below-average offers. A complex balance that is constantly being renegotiated.

It is a well-known fact that top-luxury items make ever sharper price jumps, which are not necessarily solely related to more expensive production conditions. A Gucci handbag costs several times as much as a mass-produced handbag, but is perhaps only slightly more expensive to produce and uses components that may also be part of a cheap handbag or even be produced in the same factory. So you not only buy slightly better quality, but also a name and thus also a kind of storytelling. The Gucci name itself becomes a crucial part of the selling price – just being able to afford the bag increases my social status, it doesn’t really matter that it’s really made with only a few more expenses than the average handbag.

I am telling all of this because I suspect that in the competitive market changed by AI, similar processes could take place in music. There will be forms of mass-produced goods that can be produced/composed by AIs much cheaper and faster than a human being can ever do. To be able to compete with a human production in exactly this segment will be very difficult if not impossible. This can already be seen in the stock market evaluation of Universal Music, for example, which was immediately downgraded from “outperform” to “underperform” when the topic of AI music came up.

But there will also be „special brands“, artistic personalities with which the „consumers“ (paying listeners or clients) are either linked by a real or possibly hyped type of storytelling. The storytelling creates a certain status that is worth extra money.

If the music mass market will be mainly AI-generated, clients will continue to afford human creativity as a kind of „luxury“. They might brag by name like the owner of a Gucci handbag does. The music of these „luxury composers“ may not be better than the music of the AIs, they take longer to write and have „human“ weaknesses, but this will increase their value in the eyes of their clients, because in contrast to the anonymous AIs, these creators have a name and a story that people associate with their works.

If I mention the name „John Williams“ somewhere today, numerous associations resonate just from this name. Perhaps it will be the childhood experience of seeing the first „Star Wars“ film in the cinema and being overwhelmed by the music. Tears of emotion while watching „ET“. Being scared by „Jaws“. Shocked by“Schindler’s List“. Not only does the quality of his music play a role here, but also the experiences one associates with it. You might have seen him handed an Oscar. By all accounts, he’s a nice person who has honestly earned his success through hard work, has a good craft and can look back on a long and successful career. All of this plays a role in every moment listening to his music. Last but not least, the album featuring his music might use iconic images from the films that use his music. These also create certain associations – when listening to the „Indiana Jones“ motif, we immediately see adventurous action scenes in front of our inner eye or see a world map on which dots move to exotic locations.

Now imagine the following experiment: a subject is played two pieces of music, both of which were composed by an AI. Before the first piece, she is told that this is an AI piece that was automatically generated by an algorithm. Before the second track, she is told that this is a piece by John Williams that he wrote for a film that you don’t know. The test person should then judge which music moved her more.

The next subject is told exactly the opposite: the first piece is said to be by Williams, the second by an AI. I am sure that the result of this experiment would always be that the piece allegedly created by Williams sounds more „sympathetic“ because it is associated with a living person and has more assumed quality. Conversely, the experiment would also work – music that was actually written by Williams but was unknown to the subjects would be rated worse because it was assumed to be „mass-produced“.

This somewhat unfair experiment is no proof that Williams‘ music and the AI’s music are absolutely the same. Certainly the music of the real Williams is closer to human experience and desire. Maybe his music is „more artistically valuable“ in ways we don’t fully understand today, but maybe in 20 years, when AIs become commonplace. But we shouldn’t delude ourselves into believing that these are really extreme quality differences, remember the Gucci handbag? And the AIs keep learning because we’re judging their work and constantly telling them what we like best.

ChatGPT is still writing unspeakably boring and average texts. But every time we share a text from ChatGPT, the algorithms can learn which texts perform better and which perform worse simply by how many times that text is shared. It learns what we find funny and what we find boring. And it keeps getting better at writing something funny, even if current attempts are mostly pathetic.

In this very interesting and detailed interviewwith the CEO of ChatGPT, it becomes clear that the inventors of this AI have long since lost understanding of how exactly ChatGPT learns. You can only make vague predictions about the direction in which the program will develop. So it’s only a matter of time before ChatGPT’s latest version really surprises us with something that we find creative/absurd/extraordinary and that isn’t just a knockoff of what we already have. And that’s the point where we’ll only get a higher „value“ of the artistic product with storytelling. At least until  AI personalities suddenly emerge with which we associate certain stories and images (which has already partially happened in K-Pop and elsewhere).

The „humanization“ of robots and artificial beings has not only been a topic since „Metropolis“ and „Frankenstein“. C3PO and R2D2 in Star Wars or Data and the „Holographic Doctor“ in Star Trek are not minor side chracters, but bearers of emotions and  therefore important characters that we love and adore. We want them to become „human“ just like we want poor Pinocchio to become a real boy. Likewise, it could be that there will be „AI personalities“ with names and stories that represent a different „style“ and even act in direct competition with human personalities. Perhaps “brand awareness” will also develop with algorithms. Do I use Microsoft’s „average“ AI or Apple’s „premium“ AI? The users will decide.

How will we humans deal with this new competition? Quite simply: we will do what we are actually very good at: we will cheat.

More on that in the next article!


Moritz Eggert


Liste(n) auswählen:
Unsere Newsletter informieren Sie über Neuigkeiten im Badblog Of Musick. Informationen zum Anmeldeverfahren, Versanddienstleister, statistischer Auswertung und Widerruf finden Sie in unserer Datenschutzbestimmungen.
 | Website