Raw LLM Responses

9:10 I have to disagree with you on "tracing this is what AI is doing". There are some automated tools that do procedural clip art, that is in fact taking bits of existing images and combine them together. But thise are usually not AI directed. They can use AI to fill in the gaps and hide the seams, but are not representative of how generative AI works. The actual AI is frighteningly similar to how a human would learn the Art. They examine existing peices and associating them with descriptors. The internal data in AI does not memorise the image. It is just a guide set of numbers that are used to calculate each pixel of an output based on set of descriptor flags in the input. If you only took one picture - sure the internal data will be basically that picture. But if you give several pictures to train it will turn into a blur or all the pictures and the more images you add the less each specific one gets represented in the internal data and instead it will start finding out commonalities and abstract concepts that are shared by some but diffirintiate from others. Like a human would with more exposure get more experience and gain more understanding of deeper concepts and not just surface level descriptions. And so when a generative AI is triggered it gets some input - be it prompt or image data from surrounding areas - some random noise and based on those values it does some simple math to generate a pixel map for you and it so happens that with enough network complexity from this basic math emerges something passable as art. Not because the AI did a search in it's reference library and cobbled something together... But all that being said - all your previous objections - corporate greed using AI to push beginner artists out of jobs, unethical use in propaganda or NSFW, using people unique styles without their consent and in spite of their objections - all those objections are still valid and I totally stand with you on that one. And the main reason why I feel necessary to correct you on that statement about how AI works is because that statement being incorrect can compromise the rest of your critique of AI malpractice. Like "if you are not even aware how AI actually works why should people take your critique of AI seriously" arguments. Besides I'd just taking snippets of art was all there was to AI this poisonin would not have been effective.