Raw LLM Responses

​​​@gehrtz A world where people are spoonfed hyper-personalized AI-generated content to keep them scrolling and watching ads doesn't sound good to me. This benefits platforms, not people. Artists should absolutely be able to opt out, but companies deliberately avoided that approach. For instance, Meta wanted to license some training material, but were worried about cost, time, and undercutting their fair use argument. Human creativity was absolutely essential: the words, images and videos of God knows how many millions of people are in these datasets. No scraping, no generative AI. Synthetic data came after they gobbled up human-made data and it isn't as good. That's why these companies want exemptions to copyright and why OpenAI wants to start a social media company. The human data is better and they depend on it. Human learning cannot be reduced down to pattern recognition. We don't fully understand the human brain, so we can't truly say that LLMs learn like humans do. LLMs sometimes fail to follow simple instructions that even a child could execute. They're next-token predictors, not brains. They have no reasoning, logic, or ability to understand the world, unlike humans. It's a common myth that artists are somehow in it for the money. The top of the art world might have pieces selling for millions, but most of us do art on the side. Though reselling at auction is where you see those eye-popping figures, and the artists don't see a dime of it. Most of us just love to create and would like to make at least part of our living doing so. That's what people do: they develop a skill that they like and they monetize it in order to support themselves. This doesn't make artists or anyone else greedy.