This is partially a repost of my comment in the news megathread but not really.
OpenAI just announced Sora, a tool for creating video from text, and the results are really fucking good (especially compared to state-of-the-art AI video generation tools), and this has me thinking about job security again.
Generative AI is already displacing workers.
A study surveying 300 leaders across the entertainment industry reports that three-fourths of respondents indicated that AI tools supported the elimination, reduction or consolidation of jobs at their companies. Over the next three years, it estimates that nearly 204,000 positions will be adversely affected.
The Concept Art Assn. and the Animation Guild commissioned the report, which was conducted from Nov. 17 to Dec. 22 by consulting firm CVL Economics, amid concerns from members over the impact of AI on their work. Among the issues is that concept artists are increasingly being asked to “clean up” AI-generated works by studios, lowering their billed hours and the pool of available jobs, says Nicole Hendrix, founder of the advocacy group.
“We’re seeing a lot of role consolidation and reduction,” Hendrix says. “A lot of people are out of work right now.”
According to the report, nearly 77 percent of respondents use AI image generators enabling, for example, individuals to upload landscape photos to virtual productions screens or speed up rotoscoping in postproduction. They have applications in 3D modeling, storyboarding, animation and concept art, among other things.
Generative AI displacing workers isn't some future hypothetical, it's something that's already happening right now, and as someone working in a field which is vulnerable to automation by AI tools, I'm really worried that OpenAI (or some other company) is going to create a new tool that just completely puts me out of a job.
Is anyone else worried for their job? Is there anything that can be done?
No, it's largely a tech bro grift similar to cyber trucks and crypto. Yes, the bourgeoisie will fall for it and try to implement it (See Amazon's embarrassing AI product descriptions) but capitalism is already crashing and the job losses while replacing stuff with a terrible facsimile will just crash it harder. There will be an AI crash as it infects everything and makes actually making accurate predictions of reality impossible. Scientific research, for example, will be fucked by inaccurate AI models. This will include engineering and tech.
No one actually likes this shit. Something will give eventually. It's not even actual AI, just an algorithm.
the hip new trend for the late 20s is going to be forgetting that the Internet ever existed
the fact that the Internet is becoming completely unusable, an infinite grey sludge landfill, is going to make this easier than it feels at the moment
We'll probably settle on 2001 where the internet is just for weirdos and normal people only use it to get directions and check their bank account
I mean, "AI will crash eventually" is little relief when animators are getting put out of jobs right now. Everyone can talk all they want about how generative AI can't produce real art, but it's clear that AI is already good enough and getting way better each year. I'm probably not going to get put out of a job in 2024, but what about 2025 or 2026?
Oh, totally. I didn't mean it wouldn't suck before it gets better. It's horrible.
Like a part of me dies knowing video games that are lazy asset flips with AI-assisted character designs are being eaten up by consumers.
Capitalism has spent decades convincing people not to value artists, writers and scientists and it's really fucking sad to see some people's attitudes being "Well writers have been making lazy shit lately so I don't care if they get replaced". Mother fucker those writers aren't lazy, they're working in terrible conditions and have to have their work butchered by marketing teams.
People are going to notice though, people already hate Marvel movies, but with AI shit the bourgeoisie will have even more control over art, and the quality will be even worse than what people hate now. It's popular for now because it's new, but people are already kind of starting to dislike it for its repetitive nature and soullessness.
A few years ago, when artists were speaking out against generative AI, I laughed at them because I never thought that AI was going to be able to draw hands and that any artist so terrible they could be threatened by AI that can't draw hands deserved their misery.
I was wrong to think that way, because a year later, AI figured out how to draw hands. And today, my own work is seriously threatened by generative AI.
AI allows mediocre people to make an endless stream of mediocre, dull, aseptic, sterile content.
I suspect that there will be so much terrible shit made with AI that they wind up having to hire people to correct it.
I kinda think this Sora thing is gonna crash hard once people make revenge porn with it.
I mean, artists are already being asked to "clean up" AI-generated images, so this is already happening lol.
This already happened with machine translation. Professionals get a pay cut, profits increase, more people get access to mediocre or bad translation, yet good translation does not become more commonplace. So likewise I expect mass layoffs of artists and an even higher ratio of terrible movies/shows/movies to good ones.
How long can you really say this is the case? LLMs and diffusion models in 2024 are so much better in terms of capability than conventional algorithms that it's just a truism to say "it's just an algorithm".
Like, technically, lions and tigers are just biochemical reactions and atoms, but if I'm getting chased by one, knowing that isn't going to bring me any solace.
It only took a few years from shitty dall-e mini memes to photorealistic images. Even if AI isn't "good enough" (a standard that keeps on getting raised every time a new development comes out), what about in 2025? What about in 2026?
Like, "AI can't draw hands" used to be a meme, but AI can draw hands now. AI used to not be able to form coherent backgrounds, but AI can form coherent and consistent 3D environments now. I think people have a right to be worried!
GPT used to not be able to remember long conversations, but Gemini 1.5 can remember 1 million tokens now.
This technology is getting better day by day, and it's showing no signs of stopping.