That’s probably the clearest statement of the dangers AI poses for just about every field.
1 Like
composing
I was unsure whether this was an AI aberration, a typo, or some subtle observation about the creative process.
5 Likes
It’s a work of art. And you, as part of the artistic microcosm, get to decide what it was.
BTW I really like that kind of humour…
2 Likes
Ah, dear colleagues…
This endless “pros” and “cons” about the “AI” vs “Real life”.
I would share my opinion here.
The AI could be very helpful in the music area. It could help us getting decent sounding audio output out of our Notation apps, or DAWs.
We could use them for providing demo versions, or even nice sounding music for multimedia and theaters.
On the other side, the AI won’t be able to replace the real symphonic orchestra in the concert halls, or for bigger multimedia projects.
The AI has it’s place in the music world, and it won’t replace the live musicians. 
The only genres that could be heavily affected by the AI, are those pop genres which, predominantly relate on virtual synthesizers and drum machines - House, RnB, Trap, Dubstep, K-pop, J-pop… etc.
Best wishes,
Thurisaz
2 Likes
That was the AI in his phone taking over his world. 
I’d been keeping out of this, but am now drinking a few drams, so decided to chime in. Slainte! 
Yep. This is changing more rapidly than most realize.
Absolutely true. Just providing a link as evidence.
Also true and happening already. All the BBL Drizzy remixes were some of the biggest hits of last summer. The original was entirely AI, although the prompts were provided by an expert AI Prompt Engineer wiz who also just happens to be hilarious:
I think it will replace a lot of demand for orchestral musicians unfortunately. There’s just too much of an orchestral ideal in terms of sound that is more easily mimicable. I personally can tell the difference between Rampal, Galway, and Pahud playing the Ibert, but to 99.99% of the public it’s just “flute.” Once we get past a personal interpretive level, there’s a group cohesive ideal that can actually be approached by an AI instrumental section.
Fortunately for me (as primarily a jazz saxophonist) no AI can respond in real time (yet), and jazz saxophone sounds are too individualized for there to be a cohesive ideal, so there’s not as much of a chance of AI replacing a live jazz performance anytime soon. Some of the best VST manufacturers are essentially posting garbage as demonstrations because even they can’t do better. I generally love VSL stuff, but they posted this emulation of a terrible 9th grader thinking it sounded good. I mean, this is with substantial human editing, and we’re not even remotely close to being in the ballpark of a professional performance (I spared you the worst of it):
Better AI renderings are on the way though. There will be a day sometime in the near future when I will be able to upload a Dorico file to some AI and say play this sax section in the style of Ellington’s section of Hodges, Procope, Hamilton, Gonsalves, and Carney, and it will be able to do a decent replication. It’s coming.
Here’s a Bloomberg gift link from today regarding the effects of AI on recent college grads. My daughter was fortunate enough to get a really good internship in NYC this summer, but quite a few of her peers missed out, as internships were exceeding hard to come by this year. A lot of jobs have simply been replaced by AI. All the kids that believed the “learn to code” mantra are pretty hosed, as there are few entry level coding jobs to be had. Even excellent schools that pivoted towards “skills-based” learning instead of critical thinking are having trouble placing students now.
Critical thinking skills (and professor connections) are really the most valuable tools for recent grads. If you can think and adapt, you aren’t at risk of being replaced by AI.
7 Likes
Earlier this spring I was talking with a writer friend who is following AI very closely. When I mentioned to him I’d spent a bunch of time inputting my favorite Stravinsky scores (among others), he said this:
Before long, you’ll be able to write out something by hand, show it to your computer’s camera, and it will appear as an immaculately notated score, with beautiful playback. And you’ll be able to say, let’s try that flute line in the clarinet instead, and the computer will do it.
Pretty thought provoking.
4 Likes
That’s coming, no doubt, but I think the score recognization aspect is still quite a ways off. My hunch is the playback will be there before the scanning.
Honestly, one of the elements that fascinates me most about all of this is how much the proliferation of AI Slop is going to sabotage AI. As AI trains itself more and more on crappy AI Slop reproductions, how much does it move away from human rendering, and does it ever recover back in that direction? An AI that has trained on crummy AI Slop renderings instead of actual orchestral recordings is simply never going to approach an actual orchestra.
6 Likes
It’s very simple: dynamics, agogics, and all the markings written in the score (except, generally, slurs). And this is done via prompt only where necessary, while everything else is left to the AI. Exactly like NotePerformer does now, but using generative AI and with the possibility of controlling any other parameter via prompt.
I agree with everything you say except that. An integral part of electronic music is the (live) shaping of the sound. The very slow transition from one sound and feel to another. I’m not a electronic musician but I know that there is a heavy misunderstanding of how it works.
The only industry that will be affected is the sample industry and people that earn money by making mockups for composers.
1 Like
While I like that thought I can only imagine doing that is for sketching out something while sitting on the piano. Why would I do something that takes me much longer and needs specialised paper and a pen and a rubber when I could use Dorico on the MacBook or iPad?
1 Like
Oh! Glad to know that Fred 
2 Likes
I agree here, for the most part. I love digging into the details to make things better. However, what I want most is for humans to discover and want to perform my music, and the best way (for me at this point) is to have the best mockup I can deliver. What I don’t want to do is spend time on creating that. I’d rather compose, get that mockup out as quickly as possible, and move on to the next composition. I write almost entirely on spec, and the mockup is what I use as a demo when I put my music up for sale.
I think a lot of us are in this boat…
1 Like
A knife is a tool. A hammer is a tool…
Interesting discussion, particularly in observing how the members of this community bring such varied musical backgrounds, perspectives, and, well, degrees of civilization! Hats off to the team behind Dorico, who not only grapple with the complexities of musical notation but also navigate the expectations and demands of these diverse approaches.
Here is how it goes in my world:
AI as a player - completely useless. We don’t rely on mockups, and in fact, we prefer not to listen to recordings of the pieces we’re about to perform. The unique personality of a musician is what truly makes the difference—more so than even their technical skill;
AI as a composer – perhaps one day, but certainly not with the current approach. Most of the works that are worth performing/listening were quite improbable and deemed “wrong” when first heard.
On a more personal level, I find that humans who seek “the correct result” are far more detrimental to music than any AI I’ve ever encountered.
7 Likes