The infancy period of this technology is fascinating. Think about computer graphics 15 years ago. Beowulf came out in 2007, and was developed in the preceding

The infancy period of this technology is fascinating. Think about computer graphics 15 years ago. Beowulf came out in 2007, and was developed in the preceding years- let's call it 15+ years old. And it was right there in the uncanny valley where it didn't look real, but it looks realistic. It was interesting visually, but my brain told me "this isn't correct". And now some modern game engines are doing more realistic rendering than that in real-time. Now look at these generative models. Some state of the art ones with humans helping are pretty convincing, but it's slow work. The more general ones like these are making these wonderfully interesting images that our brains immediately say "That's not correct". But where will this technology be in another 15 years? I think the possibilities for entertainment are really interesting. Imagine a D&D game where the GM is vocally telling the AI what to generate, then making small tweaks, and the players are seeing the results.