471–480 of 579 posts
I would like to point out that there is not as much of an uproar on this forum over DALL-E utilizing other people’s photographs, illustrations, et al, as there is around Copilot utilizing other people’s code. Why do you think that is?
- Unlike software photography/illustration are not used to run essential systems like banking, manufacturing, medical equipment etc
Slightly offtopic, but how one would report false-positive check in content policy check?
I was supposed to be making a video game, but got a bit sidetracked when DALL·E came out and made this website on the side: http://dailywrong.com/ (yes I should get SSL). It's like The Onion, but all the articles are made with GPT-3 and DALL·E. I start with an interesting DALL·E image, then describe it to GPT-3 and ask it for an Onion-like article on the topic. The results are surprisingly good.
The results with things that are artworks or more general concepts are fascinating, but there is for sure something creepy with "photorealistic" human eyes and faces going on... If you want to see some really creepy AI generated human "photo" faces, take a look at Bots of New York: https://www.facebook.com/botsofnewyork
I was supposed to be making a video game, but got a bit sidetracked when DALL·E came out and made this website on the side: http://dailywrong.com/ (yes I should get SSL). It's like The Onion, but all the articles are made with GPT-3 and DALL·E. I start with an interesting DALL·E image, then describe it to GPT-3 and ask it for an Onion-like article on the topic. The results are surprisingly good.
Earlier quoted context omitted.
> I think people don't realize how huge these models really are. They really aren't that large by the contemporary scaling race standards. DALLE-2 has 3.5B parameters, which should fit on an old GPU like Nvidia RTX2080, especially if you optimize your model for inference [1][2] which is commonly done by ML engineers to minimize costs. With optimized model, your memory footprint is ~1 byte per parameter, and some less…
It's true that image models are much less of a burden on GPU VRAM than a model like BLOOM where fitting it into a few A100s is ideal, but these diffusion models are a PITA for a ordinary hobbyist in terms of total compute: the CLIP pass over the text input is almost free, but then you feed it into the diffusion model, for one sample you'll be doing 10-100 forward passes (depending on how fancy the diffusion methods a…
Excet we’re already waiting 90+ seconds for dall-e mini.
Earlier quoted context omitted.
Isn't every "new style" just a novel combination of pre-existing concepts? Nothing new under the sun and all that. Either way, I feel like your view is an exhaustingly pessimistic take on AI-generated art. I mean, sure, most of what DALL-E generates is pretty mundane, but other times I have been surprised at how bizarre and unique certain images are. You seem to imply that because an AI is not human, its art is not i…
> Isn't every "new style" just a novel combination of pre-existing concepts? At the extreme limit, maybe. But within art or even digital art, then new styles are actually not that rare, humans are pretty good at generating them. Maybe they grab inspiration from nature, visual phenomena, etc, so in that sense it's not "new" but it is "new to the medium". In art you new styles all the time. DALL-E will never do that by…
Earlier quoted context omitted.
> Isn't every "new style" just a novel combination of pre-existing concepts? At the extreme limit, maybe. But within art or even digital art, then new styles are actually not that rare, humans are pretty good at generating them. Maybe they grab inspiration from nature, visual phenomena, etc, so in that sense it's not "new" but it is "new to the medium". In art you new styles all the time. DALL-E will never do that by…
> It's very nature is without-meaning and without-originality. That's a pretty bold claim. What are humans but statistical models that generate based on pre-existing training -- and yet, are humans not with-meaning and with-originality? Of course, human brains are some large order of magnitude more complex than the neural nets that underlie most AIs, but we can already see areas where these "simplistic" AIs outperfor…
One thing humans seem to have that is beyond statistics is creativity. In that stats explain what is, and creativity takes what is and makes a dot outside that other people appreciate. No model has demonstrated even attempting the dot, let alone having a good chance of success. What DALL-E does is draw a dot between a few existing points, but never outside.
Humans incidentally have three more things that make for interestingness: emotions derived from feelings, long term memory, and roughly storytelling (~ an ability to turn long term memory into long form recall with a specific reaction intended to a specific audience). I don’t think ML has any of those, but it likely (eventually) gets the latter two.
Meaningness/interestingness require at least a few of those, and it’s what puts most art in a different category than games or math.
Wait until someone trains a model like this, for porn. There seems to be a post-DALLE obscenity detector on openAI's tool, as so far I've found it to be entirely robust against deliberate typos designed to avoid simple 'bad word lists'. Ask it for a "pruple violon" and you get purple violins... you get the deal. "Metastable" prompts that may or may not generate obscene (content with nudity, guns, violence as I've fou…
Something I haven’t seen anyone talking about with these huge models: how do future models get trained when more content online is model generated to start with? Presumably you don’t wanna train a model on autogenerated images or text, but you can’t necessarily know which is which.
I wonder if human artists can demand that their work not be used for modelling. So as the robots are stuck using older styles for their creations, the humans will keep creating new styles of art.