For Art:

Perfectly fine for placeholders for mock ups and helping convey what you want to an artist, acting as a form of interop between non-artist and artist. IE "Id like something like this", but final art assets should be real ones made by real artists.

For UI, layout, UX:

Completely fine for mocking up, much the same as art. You can rapid prototype placeholder layouts to get passed blockers and start developing, and the actual artist can catch up after.

Coding:

I work as a software dev, AI given proper tooling and setup codes perfectly fine now, but still needs human review and guidance to keep it on the rails. You still end up with human developers steering the AI, they just do their job faster

It heavily benefits from "momentum", the more good approved code in the system, the better the AI is at staying on the rails.

Theres a faction of devs that think even that can be automated away. I consider people who have gone that far to have drunken the kool-aid. I trust AI output as far as I can throw it.

Audio:

I have like zero experience on this, but every sample Ive seen comes across as very poor quality, its barely even good enough for placeholder mocks but maybe there are bespoke tools out there? I haven't seen any though yet. IE SFX and etc.

Writing: Honestly perfectly fine like coding, given once again a human is still steering it heavily and you have proper tooling to control it. However since this is "user facing" Id consider it a liability.

Also afaik AI writing is very prone to issues and being very bland, so I would not use it heavily.

Editting: However for editting I can already confirm it's splendid. Its good at sanity checking your work and catching stuff like plot holes, inconsistencies, etc that normal grammar tools don't cover.

Its also good at helping catch stuff like using terms/words/phrases too much in ways normal tools wont.

You still want a real human editor, but the AI can help with initial sanity passes to cover a bunch of the "easy" stuff.

And its very good at applying the edits. Batch fixing mistakes, etc.

DevOps: Its splendid. It can pretty quickly do things like convert requirements into technical requirements, rapidly discover negative tests cases, call out edge cases, manage tickets...

Greatly helps with keeping that stuff organized in teams.