Speaker A and B debate whether DALL·E 2, which generates images from text prompts, is actually creative or just remixing existing data. A friend who paints dismisses AI images as a database blend, while another argues the process mirrors how human brains work. The question: can computers actually be creative?
DALL-E 2, the GPT-3-enabled design system for illustrators that you can just insert some prompts and it just yields out some images, is doing what we are defining creativity pretty successfully or using existing data?
Well, I don't know if you've seen too much of this dialogue. There's some pretty heated debates. I've seen it in different conversations I've had. I was in a group chat with some friends and this came up. Someone had sent a picture where someone's like, oh, look at this piece of art I created. And one of my friends in the group chat was an artist. She was like, that's not art. And the guy's like, yeah, it's art. It's algorithmically created art. It's art. She's like, no, it's not creativity. That's just scanning a database of existing images and blending them together. And my other friend was like, yeah, but isn't that just what our brain's doing and now code is doing it instead? And it got into this really interesting kind of back and forth debate. Like, can a computer mimic the human brain?
