It would become like books: A single author creates the entire work, except it's audiovisual instead of written. You can already do that now with written text, audio (if you have a talent for voices), and animation (if you can draw), why does adding realistic video make it "scary"?
Why is that bad? It seems great to me - way way way more stuff will be available, and it will be open to anyone (the cost is high now, but it will go down of course).
A movie is not just a book with visuals. They're completely different art forms.
Although, personally, I tend to prefer live theater over modern movies, so am less attached to the movie as an art form. But it does threaten to be a loss of an art form, and that's rarely a good thing.
If actors end up being replaced with simulacra in movies, then that particular form of art is indeed lost, in my view, because there will be no more actors in movies.
The writers are also on strike.
The model here is a writer writes a movies, hires a non-union digital model (since union models are not allowed), and then produces a movie by guiding an AI until it has the look the write likes.
The entire thing is non-union by people working for themselves.
Both the Studios and the actors union are essentially writing themselves out of relevance. And if the writers union prohibits personal AI writing (unclear if they want that) then they too will be left out.
Actors could have a role here those - if they want it. Instead of providing the digital model (those would be easily made from anyone good looking) a new job for an actor would be AI guide, guiding the AI into a nice performance.
This is the model of the future, the only question is if the current unions want to be part of it.
But on the grander scale, I agree with you. It isn't scary. "AI" is poised to democratize art like we've never seen before, and if "AI" can replace human artists that points more to the brutal reality that art actually isn't as valuable as some of us want it to be.
Art has always been based on algorithms, the appeal of art is determined purely by the whims of society and the beholders without any regard for the process.
Art that does not suit the algorithms of its time do not survive the test of time.
>It's also pretty funny to see people like you talk about democratizing art, then in the next breath talk about art as if it has no value, despite that art being extremely valuable to the training of these models in the first place.
When everyone can create art, art has no value. Value is always determined by supply against demand, and art is currently poised to see an absolute oversupply in the coming century.
This is also discounting the fact that the real value of art has always been low, "Artists are destitute." isn't a stereotype for no reason. The brutal reality is art really isn't valuable in the first place (it is after all an unnecessary and inefficient luxury) and "AI" is only going to make that fact even further apparent.
>Hell, this might even make human art more valuable as the price of autogenerated, samey, computer generated sludge goes to zero as the supply explodes.
That assumes human art is somehow more valuable than "AI" art, but I don't necessarily believe that could be a constant forever.
Photographs were considered inferior to illustrations when it was first invented, likewise the invention of block printing and then typewriting to calligraphy, and look where we are now.
In this case, art that succeeds is whatever art is in demand from the people at the time and which the process involved best reflects the ultimate valuation of that art. It's a problem to be solved, aka an algorithm.
Just like how the printing press and the typewriter democratized all forms of writing, where before writing was the sole domain of trained scribes and clerics (hence the terms "inscribing", "clerical error", etc.), so too will "AI" democratize whatever art forms it can satisfactorily produce.
If "AI" ends up replacing artists and creators wholesale, that means the art products thereof were not particularly valuable in the first place. A prominent example would be tabloid and newspaper articles, which have actually been written by "AI"s for quite a long while now.
Is any of this scary? Unless you have money or fame to lose in the shifting sands of time, no it is not. If you are such a person, well you certainly have my sympathy (unless you're a journalist; that occupation is a cancer upon society), but time waits for absolutely noone.
You can certainly come up with algorithms that can result in stuff that conforms to the given restraints, but within the set of those restraints is a large amount of creativity that is no more amenable to algorithmic computation than any other creativity.
it really hasn't. the art that you see in museums are all examples of ideas and forms that broke the mold.
> When everyone can create art, art has no value
everyone _can_ create art today. art still has value.
> Value is always determined by supply against demand
if you view art as a purely capitalist enterprise. most people don't.
> The brutal reality is art really isn't valuable in the first place (it is after all an unnecessary and inefficient luxury)
art is both an expression and driver of human culture. much of our understanding of our species, our ancestors, and our lives comes from art.
It's still hilarious to me that the AI side of this argument thinks artists are gatekeeping when all you need to learn is some pens and ream of printer paper, plus a computer with an internet connection. There are literally thousands of videos on youtube from artists teaching you about art. It's maybe the most accessible it has ever been.
I _also_ think he gets the supply and demand argument very wrong. Even viewing it in terms of purely capitalistic impulse, the attention economy is a winner take all kind of deal. Some random ass holes making some trite, meaningless art using the same model will quickly drive the price of the outputs of that model to zero. There's still going to be a huge market for original work that has something to say.
Or I could use a tool like Stable Diffusion and create whatever image I want. That's what they mean by supply and demand, when it's this easy, the value of each subsequent image that's created decreases, simply because people will have already seen this type of image.
The algorithms don't work for themselves, they are prompted, controlled, and guided by humans, who choose which output lives, which is lost, and which is modified.
Nothing is handed over, it's just another tool.