Smothered by AI fakery
The assault on music, creativity and reality
When I was studying Marxist aesthetics at the University of Marxism Leninism of the Hungarian communist party, out teacher started the very first lecture with a quote that is still with me:
Music is not about the order of the sounds of the world; music is the sounding order of the world.
It is a tortured translation, that cannot even approach the elegance of the original:
“A zene nem a világ hangzórendje hanem a világ hangzó rendje.”
The point being is that music is an expression and a reflection of the orderliness of our world. It should not be seen in isolation, but in its function. A point made brilliant in its simplicity.
I am not a musician, but music is a very important part of my life, probably more than any other form of artistic expression. I cannot imagine my life without it. I am still hanging on to my vinyl records and my CDs and have a sizeable digitized library (~40K individual tracks).
Also including scores of painstakingly constructed playlists for all moods and occasions.
As Youtube is expanding into music streaming services, it is competing on engagement time. If you put on some music as background it will use its algorithms to find something to keep you engaged.
Just like on Spotify, there is an increasing library of AI generated music presented with AI generated imagery, AI bots generated comments with fake subscription. Fakery is not the main problem, the quality is. I was working on something when this came on:
At some point I had to drop what I was doing, because it was too painful to listen to it. The next few days I forced myself to listen to the full two hours trying to understand what makes it so bad. Try to see how far you can go. I cannot imagine a real musician being able to tolerate it for more than half an hour.
Like just about everything else around us, we are drowning in fakery and mediocrity. Even the language describing them is just cheap fakery:
The Most Beautiful of Classical Waltzes that You Should Listen to Once in Your Life 🍃🍃
Immerse yourself in the elegance of classical waltzes with this enchanting collection, "The Most Beautiful of Classical Waltzes that You Should Listen to Once in Your Life." Featuring timeless melodies reminiscent of Baroque masters like Mozart, Handel, and Corelli, this video transports you to a world of refined beauty. Whether you're seeking relaxation or inspiration, these graceful waltzes are sure to leave a lasting impression.”
The picture on the top and the introduction above is from this three hours+ video:
The key word is, of course, ‘reminiscent’, and naturally, it starts with minuets, not waltzes.
It would be difficult, if not impossible, to tell what is the worst part of these all-around fakeries.
The fake impressionism? The looped floating bubbles to give a hint of movement? The three-legged man on the right and Siamese twins on the left? The ballerina shoes with the stiletto heals? The total lack of understanding human anatomy? The cannibalizing and plagiarizing of musical styles and composers?
The worst is the music itself. It is zombie-music. Just dead. All the parts are there, they all seem to function, but they are not alive. The sounds have no space to live and die, no spatial reality. Every sound is on the bar. They have no breath. Their weight is uniform.
The AI makes them sound aggressive and oppressive. There is no subtleties on plucking the strings or vary the strength of the blow in the wind instruments. AI can only understand frequency and volume. It butchers tempo into strictly enforced time units. The sounds have duration, but without a beginning or an end.
Just as the pictures cannot understand human anatomy, the music cannot possibly understand and imitate the difference between a cheap violin and a Stradivarius.
As I was listening, I was trying to imagine how these models were trained. Most likely they were fed dozens, if not hundreds of performances of every one of the classical music pieces and then created an algorithmic average of them.
Western classical music is a pinnacle of Western civilization. It requires an exceptional level of cooperation, and it represents a nuanced and layered understanding of the world. It indeed represents the sounding order of the world.
The fertile ground for it was European history. The personal relationship with a monotheistic God, the emergence of civil society with reformation and the industrial revolution, but at some point it ran out of steam and started to decline.
AI didn’t cause the decline, it is only accelerating what started already in the middle of the 20th century.
Our musical taste in general is getting progressively less refined.
When it comes to popular music, the decline was already happening without it, but AI will also accelerate it. Watch this video (posted 8 years ago) for the technical details:
PJW makes the point in the end that the process is very similar to how fast-food chains engineer their products to elicit our most basic biological responses to them.
The word that came to my mind the most listening to AI classical was ‘shit’.
We could call what is happening the shitification of music. That is what algorithmic AI does. This is what happens when you average the uniqueness of live performances. It is what happens when you try to titillate your tastebuds by eating all of your favourite foods at the same time. It will give you indigestion, and it will still turn into shit in the end.
Why is this happening? Money, of course, but even more importantly, because we allow it.
Because the collective ‘we’ is willing to consume it. We could ask, how many of the brainwashed generation can tell apart the AI generated garbage from a real performance?
This performance of the William Tell overture:
from this AI generated version.
The real danger is that if we allow the AI generated garbage to crowd out the real thing, then within a generation, most of real music will be gone.
A few years ago, I was visiting a friend with teenage twins. I asked the girl what kind of music she is listening to. She showed me some manufactured pop in vogue at the time.
I showed her one of my favourites: Ojos de Brujo - Todo tiende live
From a concert recording in Greece
She couldn’t listen to it to the end. She did not like it. It’s too complicated, she said.
Forget about Bach, Wagner or Mahler, just try to guess what percentage of the Taylor Swift zombies could appreciate this relatively simple music:
… and then try to imagine an AI version of it…..
Our attention span is decreasing. Everything we consume is manipulated or engineered.
I can handle it in my personal life, I just block everything obviously AI generated.
You can do that too, but much more should be done.
Speak up and demand that all AI generated product should be labeled as such. That does not make you a luddite. AI is not bad because it is not mature enough yet, what makes it bad is the conceptual flaw of LLMs: you just cannot average creativity.
But the problem has a much deeper perspective, our understanding of what AI is and what it can be expected to do.
As I discussed in the posts linked below, AI has two possible directions: to become the best tool humanity ever had to understand and control the world, or to become an instrument of our oppression and eventual self-destruction. AI will not make the decision about its direction. One way or another, we will. Like it or not, you are part of it.
I will return to this question in my next post.
Like everything else on Substack, this is a reader supported publication.
You can help it by following or subscribing.
You can engage with it by clicking on like and/or commenting.
A ‘like’ costs nothing and is worth a lot.
You can help this Stack grow by sharing, recommending, quoting or referencing it.
You can support it by pledging your financial support.
Any and all of it will be much appreciated.
More from here
AI is no match for natural stupidity
You can find a long list of references at the end this post
References
Django Reinhardt: The Electrifying Jazz Manouche Swing
This is another AI creation of pure sh*t. It has practically nothing to do with the music of Django Reinhardt.
Excellent analysis (both Zork & PJW.) I used to produce R-to-R recordings of my university chorale and instrumental concerts, mostly in a 19th century chapel on campus with excellent reverb characteristics. There was also a pipe organ in that chapel, and music depth is never as evident when listening to Toccata & Fugue in Dm at full pipe volume in a reverb chamber (I tear up just thinking about how good that sounds and FEELS). Summer before my senior year I was in Hollywood recording studios. My formative music years were with either 70's rock or classical on albums or live recording of classics and pipe organ music. The recordings I made in college are still better than ANYTHING you can hear digitally since 1980.
Once you get used to live or uncompressed, unsampled recordings (e.g., the sound path is analog all the way), ***nothing else will do***. Even the HD quality on YouTube videos has compression and sampling and D/A conversion (and listening on a laptop or desktop without a quality sound card introduces more compression artifacts and distortion.)
Not only is music getting repetitive, banal and idiotic, high compression (drums/cymbals especially, which began in the mid '80s) eliminates high dynamic range in the recordings. Sampled (digital) music is usually low bit rate and MP3 encoded, making music sound like you're listening in a metal drum-every note is the same volume. Forget music having tonal or volume nuance.
Note there is a renaissance of people buying '70's & '80's hifi stereo setups now which, when playing high quality LPs (or even well-mastered chrome cassettes), gets you close to the all-analog sound.