61–70 of 196 posts
Can anyone here estimate what it costs OpenAI to generate, say, a 5 second video using Sora?
> Pay-per-use: $0.01-$0.10 per second of generated video, based on existing tools and computational demands.
https://medium.com/@christianray.drapete/openai-sora-underst...
Earlier quoted context omitted.
when does it credit/attribute who/what/when/where the data it copied from?
You don't need to credit for a transformative work. LLMs don't regurgitate like you think they do.
"Copilot has been found to regurgitate long sections of licensed code without providing credit — prompting this lawsuit that accuses the companies of violating copyright law on a massive scale."
https://www.theverge.com/2022/11/8/23446821/microsoft-openai...
Earlier quoted context omitted.
I’m not sure I agree it’s hard to create specific things. The tooling, imo, feels infant, but promising. And it can layer upon itself. Right now control net can do pretty good character pose control, but you have to make the pose. Well… next step is to generate the pose with a request. Same for faces.
When you're editing in an NLE, you can pick the exact frame where your transition begins. When you're editing in DAW, you can change the volume by tiny fractions of a decibel, for a single frequency. Just about every piece of mainstream media (e.g., excluding social media) has had every single frame, and every second of sound, polished to absolute perfection (or at least one sound engineer or editors perception of it…
If you’re imagining that AI tools mean it has to be a straight text -> output model that’s not correct. They are very much iterative transformations.
$800mm = a lot of jobs
Earlier quoted context omitted.
What's the market for a talented saxophone player in a major metropolitan area? It's not zero, but thanks to Spotify, it's not as big as before the advent of recorded music. Recorded music didn't just up the game, it completely changed it. Now you have ultra mega stars that make economic impacts to an area when she tours, but the market for a talented saxophone player in a major metropolitan area just isn't what it o…
Er, Taylor Swift? Your analogy is wrong, Spotify, or really the phonograph, didn't put musicians out of work. The television did, because once that came along "just" playing well wasn't enough to be a star, on camera presence became more important.
But I also don’t think the life of the musician is all that dire, at least compared to the portrait artist and tunnel digger. There are plenty of jazz clubs in any major city, and live music that pays enough to at least supplement a day job is a thing. Even at the height of live music your average musician would have needed a day job to get by.
First, labor laws isn’t the answer to technology. It would be like regulating the camera because it put portrait artists out of work, or steam drills because it puts tunnel diggers. Art improved dramatically in a short period of time with the advent of the camera, became reflective of what the camera couldn’t do - reflect the nature of humanity back to us. Second, if you’ve done much with generative imagery you know…
What's the market for a talented saxophone player in a major metropolitan area? It's not zero, but thanks to Spotify, it's not as big as before the advent of recorded music. Recorded music didn't just up the game, it completely changed it. Now you have ultra mega stars that make economic impacts to an area when she tours, but the market for a talented saxophone player in a major metropolitan area just isn't what it o…
First, labor laws isn’t the answer to technology. It would be like regulating the camera because it put portrait artists out of work, or steam drills because it puts tunnel diggers. Art improved dramatically in a short period of time with the advent of the camera, became reflective of what the camera couldn’t do - reflect the nature of humanity back to us. Second, if you’ve done much with generative imagery you know…
I disagree, we are clearly in the end phase. Generative AI (not necessarily LLM) is clearly in a completely different category from the things you describe. Suppose it becomes such that it’s equivalent to a human using all the tools of today, but instantly. What would there be to do? I guess we still need to mine for the natural materials, so there’s that.
I will wager you we will see a dramatic realignment in the sorts of work that go into production, but the demand for those skills will be so high and the value so immense that we will see a profusion of new work, better output, and more employment in the arts.
Can anyone here estimate what it costs OpenAI to generate, say, a 5 second video using Sora?
At a guess, less than $0.50? > Pay-per-use: $0.01-$0.10 per second of generated video, based on existing tools and computational demands. https://medium.com/@christianray.drapete/openai-sora-underst...
Earlier quoted context omitted.
I disagree, we are clearly in the end phase. Generative AI (not necessarily LLM) is clearly in a completely different category from the things you describe. Suppose it becomes such that it’s equivalent to a human using all the tools of today, but instantly. What would there be to do? I guess we still need to mine for the natural materials, so there’s that.
Portrait painters and tunnel diggers said the same things. Impressionism, expressionism, modern art, etc weren’t obvious outcomes. The thing AI lacks, and will lack for the foreseeable future as we have a much weaker grasp on its mechanisms, is agency and at the root of that creativity. Most movie making today is a lot of mechanical craftsmanship and a lot of that will probably disappear. The bar for good will be rai…
Most of white collar work is very much rooted in the “mechanical” tasks.