I don't think anyone is saying that they won't be able to make the technology work. This demo is impressive in lots of ways.
The question is, and what a lot of folks who are bullish on AI seem continue to ignore, what happens when the VC money runs out, who's going pay for this technology to run. You're pretending like it isn't ungodly expensive to create and mantain. There must be a product at the end of the day that many people will pay good money for. And that money has to be multiples more than it costs to run or the companies go bye-bye.
Now, this is how I see it playing out until the technology becomes cheaper and cheaper and then any/every company will have some AI engineers on staff and it will be integrated in every product everywhere. And then even lowly developers like me and my tiny sample library company will have access to the tools and will improve our libraries with it. This vision of a 'changing of the guard' and all the sample library companies get swapped out for AI instrument companies is just silly. The tech is too expensive and the profits are too small now. Once it's cheap enough to develop and run these tools/models on the new Macbook Pro, not accessing some massive server farm in the desert via the cloud; actually running on your new PC, that's when it truly permeates everything.
So, I don't think anyone can deny that eventually AI will be integrated, in some way, in all the products we use. The dollars and cents just don't work for it to happen overnight, in a market as small as ours.
I think you underestimate how much money this will generate, and how much better AI will get and be capable of generating even more money. Everytime someone has said “well it can’t do this and that” tends to be proven wrong in about 6 months or less sometimes.
We’re so used to technology taking 5 -10 years to develop to do what AI does in 6 months. Look at video games for example. Look at basic video games like Pong or space invaders to where we are now. That took decades. Now find out when Midjourney started, and look at the evolution of quality from version 1-6. Look at those first AI videos like the Will Smith eating spaghetti one and then look at Sora and see long that took.
And who cares if some companies can’t keep up and have to shut down?
That will merely mean that another company is doing it better.
If it takes so much money to generate content then if they’re not taking in enough it means it’s not being used, which means it will be cheaper.
The amount of money used to make sample libraries that are only sold once is insane. How are they still making money? How much have Spitfire spent on recording expensive orchestras over and over in expensive recording studios over and over and over and over again? Why would anyone think there’s enough money in the industry for that? In a sea of other really good companies doing the same thing no less. The business model for AI is often paying subscriptions or for credits, so they get constant streams of cash that only stops if there’s no one around to use their GPU’s that cost the money.
What might be more likely is that companies like Udio will have to compete with open source technology you’ll be able to either run on your own system, like stable diffusion, or third parties that will be able to far more easily compete with the Udio’s and Suno’s.
As for AI in general. Same deal applies, only it’s even more obvious they’re not going anywhere. Technology hasn’t ever had the speed of development that AI has, and it’s only going to get faster because we’re developing better AI chips and designing AI super intelligence that can help solve the problems that would have slowed us down.
OpenAI alone disrupts everything. Robotics is only just starting, which also involves AI.
There’s too much development and value created to just run out of customers. I really don’t think you know what is actually going on, what the potential is, what’s being worked on, and what uses there’s going to be and how rapidly expanding those use cases (and so customers) that will open up. The fact that they’re literally restricting/dumbing down their models actually ate to the public should tell you a lot.
And Amazon was able to survive for how long without even making a profit? Over a decade maybe? And from the ground up. We’re talking about companies that are already the biggest in the world with a ton of cash, they’d be able to survive for a LOT longer.
locusmag.com
www.udio.com
www.fxguide.com
