They prefer to work harder and not smarter. Forever hill climbing to nowhere.
I've never worked on a complicated codebase that started out that way until the rest of the business concerns and office politics came into effect. People may not like it, but the bureaucracy is far and away more valuable than the core functionality.
Mature codebases are years of people thinking of all the possible gotchas while solving their acute pain points. This is not fluff, but the living and breathing part of it. Without that code, it's just a machine barely doing stuff in the most obtuse ways possible that nobody wants to pay for.
I would argue that they're putting LLMs to work on that finer detail stuff, but AI is still far too dumb. No, what they're doing is playing with their skinner box.
I think culture moves a lot faster than you believe.
The broader discussion about AI and model capabilities died a couple of years ago precisely because it's so underwhelming now. People did adapt. Startups stopped hiring just to get to MVP. Coding sweatshops had huge layoffs and stopped overhiring. The corporate world got better tools for collaborations and meetings. Accessibility tools are still bad, but improving. I would argue that the a11y topic is still very ripe to be the next big thing as it continues to converge with better UI/UX instead of being an afterthought.
The layperson and tech professional alike otherwise agreed that this is a vehicle for blame game, grift, disinformation, etc. This is where all the pushback is and the topic at hand. People aren't dumb. The only people worried about "AI" are the ones who bet too big on it.
The comparison is pretty accurate though. The moment anyone dared to stray from the bootstrap defaults is when the whole thing would go to shit.
Every steaming pile said less about the development effort and so much more about the project management. This same boneheaded top-down approach is why AI isn't working for anyone without being willing to pour as much effort into babysitting as just writing the damn code yourself.
Old adages continue to ring true and as loud as ever. There's no such thing as a free lunch.
Coffee shops aren't usually a "scene" like a bar can be.
Bartenders are often the only adult in the room acting like one. When was the last time you saw fighting, crying, fainting, hit and runs, etc. on the regular at a Starbucks? Bars can be like this in even the nicest neighborhoods.
Have you ever worked at a bar or been a regular at one?
Too much or too little security will discourage customers from staying or coming back. The bartender is running the whole circus. It's nothing like serving coffee.
I dislike the donation button for a totally different reason.
As soon as I see "buy me a coffee" I get the sense that there is some kind of community, fandom, or other groupthink or ulterior motive potentially tainting the honesty of what I'm reading.
It may not be a conscious bias that the blogger has, but it makes me curious. If their blog is too similar to the other people they associate with, their credibility drops a bit. If I dig enough and find their little cult, I'm out. If I wanted to read that shit I can just go to reddit.
Not that I've looked into it much, but a thought just occurred to me. Why don't we use AI to generate lofi samples for tracker music? Why aren't there trackers with that feature bolted on? I should be able to search for bespoke and unique sounds out of thin air.
Surely that should be a very modest goal to achieve?
(re: downvotes... I say "AI" as a synthesis method, not as a way to interfere with the creative process, but I guess I have to resign myself to the fact many downvoters might be ignorant of how these musical sausages are usually made)
You can't mention "AI" and music together without a negative response. I think entrenched business interests as well as musicians and other music industry adjacent people, plus the well intentioned but poorly informed public have developed a visceral knee jerk reaction to the concept. Some of this is understandable, but I think it is mostly fueled by astroturfing.
I am a musician as well as a technology enthusiast, and I think this a very exciting time!
To respond more directly to your point than your aside, there are a smattering of models out there that can take descriptions of sounds and do a decent job of creating them. (Stable Audio 3 just released last month and can do this, for example). - I don't find them to be useful for sampling, though. I'm still much quicker dialing in a sound with knobs or sliders than a text box.
Diffusion models in music making are not going away, though. This is (at least in part) the future.
For a taste, look at some of the interesting things being done over in the Demon project - https://github.com/daydreamlive/DEMON - to me, this is a much more positive use of the tech than "type words/get song".
For your "search for bespoke and unique sounds", that sounds like a service Waves runs, called Illugen. It isn't built in to a tracker, but you can use it to generate the samples that you import into the tracker. Honestly I've never used it, I have too many samples anyway, but here it is: https://www.waves.com/illugen
But you also don't have to go down the generative / diffusion path. You can ask your AI to make a tool to generate various sample files for you mathematically. A good frontier model will happily make a program that will create some hihat and cymbal sounds out of white noise, a kick drum out of distorted sine waves. It will create some simple synth sounds from square waves. If you go deep enough, it will happily go down the rabbit hole with you into additive synthesis while nerding out about the Synclavier. Or it will do synthesis through frequency modulation while chatting about the Yamaha DX7 and don't do that because FM is like modular you'll never find a way out of that hole has anyone seen eno lately.
If you kept reading this far, there's plenty of Claudes already doing their own very simple synthesis & music writing, in videos like these below (two I found today). The models are capable of much much more if you give them time to build a toolset, and aren't just asking them to one shot an entire video about their existence using only a copy of ffmpeg:
Not quite what you suggested, but I did some experiments several months ago "enhancing" the samples in tracker music with some models, and they sounded terrible. There really is something about the sound of tracker files that's just right. But sure, you could generate lo-fi samples, there's a lot of computer generated samples in music, but putting them together into a pleasing combination is the hard bit.
Could you say more? I don't really follow, and I've used trackers for a long time. Don't some trackers already have something akin to this in terms of "randomizing" wave forms inside some reasonable parameters? Why would you need AI for this problem?
>Why don't we use AI to generate lofi samples for tracker music?
Because generating lofi samples is already pretty easy with waveform generators and existing tools. Burning millions of tokens worth of compute just to make a bass kick is profoundly wasteful.
The target codebase cannot improve beyond the point that the reports are incorrect and a waste of money.
There is also the question of whether humans can waste so much time reviewing AI code that the vulnerability is not patched before it is exploited. Another one is whether when the human is removed from the loop that the codebase becomes more vulnerable in some other ways.
Making critical decisions without oversight is just as bad, or maybe worse.
If you frame it this way in a meeting, you will get the attention you want. Don't say I didn't warn you because that comes with a lot of scrutiny you might not want.
I've never worked on a complicated codebase that started out that way until the rest of the business concerns and office politics came into effect. People may not like it, but the bureaucracy is far and away more valuable than the core functionality.
Mature codebases are years of people thinking of all the possible gotchas while solving their acute pain points. This is not fluff, but the living and breathing part of it. Without that code, it's just a machine barely doing stuff in the most obtuse ways possible that nobody wants to pay for.
I would argue that they're putting LLMs to work on that finer detail stuff, but AI is still far too dumb. No, what they're doing is playing with their skinner box.
reply