I genuinely hate AI and the entire concept (save for the VERY few actually good AI (specifically that one image recognition AI that was trained to recognize specific cells (i think cancer? Either way it was a cool medical advancement))) but the way that AI actually WORKS is quite interesting to me.
Obviously i'm not going to make an AI or something like that, but researching how LLMs and image generators function gives you quite a good understanding on WHY exactly AI doesn't actually work for writing, art, etc.
Take, for example, an image generating AI. It's just a bunch of images with text files named the same thing (at least, from what I could tell about stable diffusion in my research). That AI connects the dots between the text files and images, and, using a noise pattern, just makes stuff up. If you have 1,000 images, all with vastly different topics, but they all have many words in their accompanying text file, and each includes "human"; the AI will know the pattern for human, and create a human (to varying degrees of accuracy).
LLMs do a very similar thing. They take patterns from many more texts (e.g. books, essays, documents) and take what seems to be logical together and just shoves it all together.
When you understand the fundamentals of both, you begin to realize that they're both ONLY pattern recognition & repitition, and that's literally it. Art, on the other hand, is pattern recognition and repetition, yes, but also actual human creativity, time, effort, material (depending on the craft) and HUMAN flaws that actually make sense and often even give nuance to the art.
In essence, AI has always been bad, but actually understanding the fundamental ways it functions makes it worse.
In even more essence: I hate AI I HATE AI I HATE AI I HATE AI I HATE AI I HATE AI I HATE AI
izzy806
2087
germansingergirl
stormborns
prguitarman
romaincrisis
annaxmalina
titsay
