Raw LLM Responses

​@alfredovalenzuela1711 "Ai doesn't produce, it merely recreates what it's trained on (as opposed to humans)" - this is where we disagree. Ask a human to paint a neovius surface. 99% of people won't even try because they don't know what a neovius surface is. The only way to teach them is to show them what it looks like, or at least describe it in detail. And because the shape is complex, one glance won't cut it. They'd need to see it multiple times, study it, and only then could they recreate it from memory. Do you see how similar that is to training an AI model? The process is less efficient for the model. It might need hundreds of images of the same thing before it can paint it well. But given enough exposure, it still develops a general, abstract representation inside its neural network, in the same way a human builds a "mental model" from what they observe. Of course, like you said, humans incorporate many more signals into the creation process, like emotions or other non-visual data. However, try to honestly answer this question: can you paint anger? And I don't mean a representation of anger. I mean anger itself - the emotion. How does it look? You might create an abstract shape that represents anger, or you might use colors you associate with anger. But ultimately, you'll end up using the visual data your brain has learned over the course of your life: colors, shapes, contrast, composition etc. In the same way, AI uses what it learned during the training process. Even though there's no emotion behind it, both humans and AI still end up creating something based on everything they've seen before