Using a simple text prompt I can immediately generate any kind of image I want. It’s not that much different than a wizard casting a spell.
Yeah, especially if Q* and other AIs can be based in mathematics so it can invent new things that actually work.
Even a simple LLM is a window into the Jungian collective unconscious at time of training, but nobody seems to talk about the philosophical implications of what we have right now.
The Age of perfect robo wives.
For normal humans yes, for those that merge not so much.
There’s websites showing how you can build technomagick circuits with batteries and leds and sigils / potentiometers. I don’t know how it works or if it does but there’s a lot on it. Lucifer Faust shamanic technomagick is a book that teaches you actual magick devices you can make I believe.
I don't think it is truly sustainable, it is trained on existing art, and if art as a medium starts dying because of it, it will just train itself on other ai art, and will start to look stranger and less accurate. This is just another "techbros think that they can program a better version on an existing thing but don't bother to understand the fundamentals of that thing."
Right now it is rough for artists, because a lot of people don't see the point in hiring an artist to draw things when they could get an AI to make a "good enough" picture for free. But that is just a fad, and it is slowly fading now as people realise they aren't magic, and if you actually want something interesting or actually accurate to what you want, you need to still speak to a human.
if people understood that they are doing statistics they would realise how using predictions as a basis for future predictions is a terrible idea
The people behind this stuff have a lot of strange ideas about AI and "the singularity." They're too far gone to have anything but terrible ideas. The people making this have forgotten that this is just a statistical algorithm and not magic. So in a sense, they really are techpriests.
I'm not sure if some of them ever knew entirely possible that the people running the companies have no understanding of the technology involved
Like any good con, they just use a bunch of fancy words to make it seem way more impressive than it is, and like any con that goes too far, they start to believe their own hype and forget that they can't actually deliver what they promised.
lol why would they knowingly use inbred training data when they can just use pre-2019 sets
You underestimate their smugness and belief that their techbro cult can create superior versions of everything in the human existence. They already use inbred training data on chatbots, they'll be using it for artbots soon enough.