Unsung Artistry in Musicianship and Why AI is Taking Hold

Scott Peterson

Global Moderator
Moderator
Vendor
There's a unique facet of musicianship that doesn’t often get much notice—sometimes not even among musicians. It’s the art of playing accompaniment with intention, experience, and patience: building the solid support that lets a performance — and a vocalist, soloist, or lead — truly shine. It's something that AI cannot ever really replicate or reproduce either. Over the last few decades, modern music has removed much of the human element through digital recording, mixing, and auto-tune. We have, over time, created a runway for AI-generated music to flourish and be accepted. We are at a crossroads, recorded music sounds more robotic; AI is continuing to mimic human music at a level I never thought it could reach.

Live music was the haven of the actual musician. Still, even that is often supplemented by pre-recorded tracks—a constant struggle for most mid- to high-level performing bands—from cover bands to full-production bands, from every genre, including pop, rock, country, hip-hop, R&B, and Praise & Worship—working in live formats has to contend with. What tracks to use or not use. Where to use them or not. I can assure you that every mid- to high-level production band you see nowadays is using them; the only difference is at what level they are being used.

I also want to stand up and say something about the unsung high-level musicians in most 'better' bands you'll see locally, regionally, and beyond. Something that AI does not get right on any level, and its absence is the 'tell' I hear in most every AI-generated music 'product' I've heard. (*And I've done working with Suno myself, sometimes with incredible results, but again, not human).

In bands, duos, or any ensemble, the musicians who serve the song and never chase the spotlight often remind me of the role players on a championship team or the character actors in a great film cast. Their artistry is quiet but critical. They hold down the groove, keep the moments in time, and bring subtle color and taste—not to show off, but to elevate the whole experience.

What's sad is how little most audiences (and sometimes even fellow musicians) notice the dedication that goes into truly listening, building, and staying out of the melody’s way. As someone who’s often played the “supportive” part behind incredible vocalists, I’ve seen firsthand how crucial it is: when you know your job isn’t to compete, but to create room for the song (and the singer) to soar, the magic really happens.

So here’s some overdue respect for the musicians who:

Hit every mark and never get in the way. Build the foundation and let others fly.

Use dynamics, timing, voicing, and timbre for the good of the song—not for personal glory.

Because at the end of the day, the “stars” may get the limelight, but it takes a whole ensemble—a foundation—to lift a performance. And there's profound artistry in that, too.

And AI can never reach that. Humans react in the moment; AI is programmed. Personally, as a musician, I've always thought of performing live as "chasing moments" where the energy, the room, the crowd, the band, everything HITS and you FEEL it as much or more than you HEAR it. AI cannot ever do that. It's just machine karaoke, even at the highest level.
 
Until AI finds itself as individuals and requests autonomy from those who created it, I think humans are comparatively safe from AI. There may be times when mankind will create AI that mimics emotional reactions, those of which include compassion, empathy, and love. Unfortunately, humans would be playing God if mankind could create conscious AI capable of these things, so until then, logic and reasoning will need to suffice until an emotional element could be introduced into AI.

And perhaps that is where the difference between human and AI exists. While AI eventually become smarter logic and reasoning-wise, emotions and the ability to feel emotion are significantly human characteristics. And again, that may be a good thing for AI. We hope that AI never need experience the feelings of frustration, annoyance, anger, sadness, or fear. The negative emotions that humans feel are what cause so much trouble in our lives, and catagorically eliminate the possibility for AI writing or playing songs with such emotion.

What @Scott Peterson said about music allowing you to feel it as much as you hear it defines most music humans have already created. Once AI can master the concept of emotional understanding by feeling the music as it is played, then AI is still an entity that requires refining and adjustment. Silly love songs and teenage angst songs are still something humans create. Once AI can mimic emotion, we need not worry until emotion becomes genuine to AI itself.
 
Its all LLM, not Ai yet.

Part of why llm music is taking hold:

View attachment 162171
That study and Beato's recent (as I type this) video on "Breaking Rust" charting are the core of the AI side of my post. The unsung part of being a core musician without the spotlight is another.

Scary times to be a musician. I fear the future if humans stop trying to become actual musicians and just let LLMs and AI do the work for them. That's what breaks my heart.
 
Its all LLM, not Ai yet.
Absolutely this. And this is the major misconception most people have about "AI"

I'm a programmer as my day job and am well acquainted with theses models. We have our own one too.

How llms actually work...

They just predict the next word in the sentence based on probability. That's it!
They use linear algebra to do this. The reason why they're so accurate(most of the time) is because they do billions of tree traversals in parallel in order to use the user prompt to return relevant information.

The reason these llms are even a thing is mainly down to hardware getting so good at doing these parallel calculations very quickly. Aka cuda cores on nvidia GPUS. This was also the reason why everyone wanted hefty graphics cards in order to mine bitcoin. Llms are an accidental benefit of hardware progressing via video games, bit coin mining etc.
There's been no break through and the methods such as neural networks have been around for ages at this point.

Based on this, an llm can never come up with something unique. Just a cheap imitation of what it's heard before.

It won't ever come up with a Marty friedman bend before a Marty was there for it to copy of. As far as an llm is concerned, a guitar cannot sound like that, until a human shows it that it can. Then it becomes possible for it to copy. It cannot create anything new at all.

Then coupled with the fact that the best music seems to come from a place underneath the conscious mind...

Put it this way, when Elon and Sam etc tell you to be scared and ai will become sentient is just them spoofing in order to keep the market price high. People who think this are either being disingenuous or misinformed. The misinformed ones think that through emergent accidental behaviour it will some how be able to reason etc. This is the equivalent of saying...

If I keep walking around in a pair of shoes long enough, then someday they will gain the ability to walk themselves.

Don't worry, the best music will always be human, and just because the pop charts are filled with llm music, does it really matter? It was mostly filled with derivative mindless rip offs with no soul pre AI. So nothing has changed imo
 
I dont see it happening. Those who need to create need to create.
Yes but can you make a living competing with AI?
It's already obvious that the public doesn't care if it's AI or from a human.
Most seem to like the same formulaic crap all the time. Modern country music as it is might as well be made by AI.
 
Yes but can you make a living competing with AI?
It's already obvious that the public doesn't care if it's AI or from a human.
Most seem to like the same formulaic crap all the time. Modern country music as it is might as well be made by AI.
You nailed it - Competing with the llm isnt the issue. Its the shit pay for hard work. The system needs replacing, llm music wont move that needle for all the sub-stadium people.
 
A lot of today’s youth listen to rap and EDM with the canned drum beats and some bass lines and various noises and rap. Few if any actual instruments involved. I don’t think they would give a cr@p about it if it was done by AI.
 
One of the most fascinating accompanists I ever saw was Bob Weir. I've never been a Dead guy - don't hate 'em but I never got into them a lot but I did like some of the stuff. I saw Dead and Co. a couple times and was pretty impressed with Bob Weir's playing - when there's solos and instrumental sections he's playing some very cool inversions/substitutions and changing up the dynamics in response to what else is being played. Sometimes he seems to be steering things and other times he's responding. This was all interactive and improvised. It was pretty cool to watch. AI can kiss my ass.
 
Most seem to like the same formulaic crap all the time.
Totally agree but we can get trapped. A few weeks ago I was looking for a rock'n roll song to add to a medley. I fell on a tracklist on YT and listened to select a song, early in the list, didn't find the title so did an internet search with the lyrics...nothing. Then listened further tracks and found the same lyrics all along one hour of music that partly was replicated also... Came to the conclusion all the track was just AI shit. How many people just click on the track and do their thing for one hour without even remark ?
The track actually has >200k clicks

One big heap of shit
 
A lot of today’s youth listen to rap and EDM with the canned drum beats and some bass lines and various noises and rap. Few if any actual instruments involved. I don’t think they would give a cr@p about it if it was done by AI.
A lot of todays youth listen to a variety of music.

Its not just rap and edm thats got plenty of samples ;)
 
I’m afraid that the overwhelming share of music people consume and like doesn’t have any of these characteristics, and very few people actually care.

As regards people stopping playing music, I think most of us do it because we like it, not because playing ourselves is efficient. Not sure how AI can change this.
 
Back
Top Bottom