[nivenkos]

Loads of places where before you'd just have amateur art / no art.

We used it for department logos, etc. - it makes sense when it's so cheap.

For it to really grow though it needs to be able to be more domain-specific, like being able to generate 2D and 3D animations, spritesheets and isometric sprites, 3D models, normal maps and textures and being able to maintain a consistent style as a context (like ChatGPT).

[filoleg]

> For it to really grow though it needs to be able to be more domain-specific, like being able to generate 2D and 3D animations, spritesheets and isometric sprites, 3D models, normal maps and textures and being able to maintain a consistent style as a context (like ChatGPT).

Heavily agreed. Just like how the smartphone revolution ended up in tons of businesses that create things around that ecosystem (from apps for many specific domains to tooling to whatever else), the current "AI" revolution will kick off creation of viable companies that apply the models in functionally practical ways to specific domains (including things you've mentioned like 3D models, textures, and tons more).

At least that's my hypothesis. The only caveat that I can think of is that the currently available "AI" would need to be actually functionally viable for those subdomains. But that's a split responsibility between the state of "AI" itself and the companies making tools for all those specific domains beijg able to actually make a good use of the models.