Raw LLM Responses

Generative AI in terms of text can be summed up as "Better Search Engine than Google" Generative AI in terms of art can be summed up as "Everything it produces looks the same" The next major step for Generative AI to make any kind of headway is for it to start generating its own data by exploring the real world. By limiting the model to only things you can find online you will eventually end up with results looking more and more the same. Midjourney for example can never produce the art in my head or the artstyle that have very little data for it to train on. You may tell Midjourney that you want to draw something in Dragon Ball style and it will produce something that looks like Dragon Ball style but which one? You may tell it Akira Toriyama style to be more specific but Akira Toriyama style evolved over the course of his career. There might not be enough data of Akira Toriyama art style for it to train on, so it might supplement it with other filler (trash) data labeled as "dragon ball style" or "Dragon ball manga style". Generative AI Art as it is now is nothing more than a fancy tool that can create a base for you to work on. Companies that are laying off artists in favor of Generative AI Art will struggle eventually when everything they are producing looks the same as everyone else.