[ExI] Spotify’s Attempt to Fight AI Slop Music Fails

[Ben Zaiboc]

LOL

"AI music provides an accessible and affordable solution to a major ... 
problem: securing royalty-free soundtracks. Platforms like Soundful and 
Beatoven.ai specifically market their services as a way to generate 
unique, royalty-free background music for videos, livestreams, and games 
at the click of a button. This convenience and cost-effectiveness appeal 
directly to creators who want to avoid copyright strikes and high 
licensing fees"

So copyright and licensing fees are directly driving this phenomenon 
that the copyright industry are so tied in knots about.
I think this is what's called Irony.



"The fact that a song is 'popular' on a technical level is no longer a 
guaranteed reflection of its creative value"

Being popular on any level has never been a reflection of 'creative 
value' (although I have to admit, I don't really know what 'creative 
value' actually means).
We all know, most popular music is drivel. AI-generated music doesn't 
change that in any way, in fact, it might even be an improvement.



"...often producing lines that sound like they were pulled from a random 
quote generator"

Cool, that's how Davie Bowie produced some of his lyrics!
But in any case, random lyrics are likely to be an improvement on most 
of the lyrics from pop songs over the last couple of decades, at least.


"AI models are trained on existing data ... A heavy reliance on AI could 
lead to a feedback loop where new art is merely a pastiche of old art"

Just think about that for a moment, and compare it to how humans have 
always created art.
As Picasso (or TS Eliot, or someone) allegedly said: Good artists copy, 
great artists steal. Creating a pastiche of known things is what 
'Creativity' actually is. There's no other way of being creative, if you 
think about it enough.


"Often relies on loops; transitions can be ... smooth or abrupt"
"Struggles with deeper meaning and emotional coherence; may sound like 
phrases from a random generator"

So, pretty much like most current dance and pop music, then.



"... avenues that AI cannot replicate, such as live performances..."

Ha!, tell that to the Koreans and Japanese! There are probably more 
places now, where digital entities give live performances (yes, it's a 
gimmick now, but the gimmick demonstrates that it can be done).



Just as with other 'AI' systems nowadays, these things are analysing and 
re-synthesising what people have already put on the internet, so it's a 
reflection of what people listen to. How can that fail to be popular and 
relevant?

The good thing about it is, the copyright parasites haven't yet figured 
out how to profit from it, so they're squealing that they are suffering 
from 'lost revenue' (i.e. money that they think they could be making but 
aren't). I'm sure that's just a matter of time, though. We can expect a 
rapid about-face when they stop trying to combat it and start trying to 
control it.

Having said that, it does seem that the space that profiteering 
middlemen live in, and the influence they can have, is getting smaller 
and smaller, which can only be a good thing for culture in general. I 
just hope that this can extend to the throttlehold that various 
publishers have on scientific publications, with their double-profit 
model (charge to publish, and also charge to read), a scandalous 
situation that occasionally gets complained about, but nobody so far 
seems to be able to remedy. I'm thinking there may be an AI solution to 
this as well.


This is what Technological Disruption looks like. There are good and bad 
parts, but overall, it's progress.

-- 
Ben