Raw LLM Responses

Having finished the video: I’m a programmer, vaguely. I love using building blocks of code words, which themselves are building blocks of subroutines, which themselves are building blocks of a bunch of tiny electrical pulses, to solve problems. When generative AI first started “being a thing”, I was fascinated by it, especially with how dreamlike some of the early images were… and then all the years since have done is disillusion me. And given me an idea that I probably don’t have anywhere near the knowledge to even TRY to experiment with. An “ai artist” that is not just an image generator with a purpose built algorithm made out of pre existing schlock. An AI brain attached to a drawing program, with a cursor and pen colors and everything. It would be trained and guided exclusively on its own previous work, and work I directly draw into the drawing program myself by my own hand. I would teach it what a “line” is, what “shapes” are, and build outward with more and more complex keywords. Everything it can do is either directly taught to it “the hard way” or organically pieced together from multiple “hard way” trained concepts. It would probably not even be all that worth the trouble, and I would only reduce the environmental impact of this thing as it “grows” by supplying it renewable energy I harvest myself, completely off the grid, which would make the whole thing all the more of a pointless logistic nightmare, but… image-theft-less algorithms could be nice, especially if they have basically only one “style” they can draw in and cannot be “mistaken” for photographs… oh well.