Docbear64 The same thing happens with AI art like Midjourney. I work in a field and own a company requiring concept artists. These guys saw the advantages of sketching with AI and then polishing by hand where needed, and I couldn't be happier with it.
The process is (or will soon be): rough AI sketching using a model trained with general data and, maybe, some "dumbed down" instructions. In VR scripts, this translates to a model trained with 100s of videos tied with script data from different scripters. It seems like a lot of data to generate great scripts (thousands of frames per vid), but it isn't. For now, it makes the "rough" patterns and can get 90% of a funscript done, but it still lacks the fine detail that comes with USING the script and tweaking the desired feeling on the fly. So, the scripter/company could release a rough script that does the job, yes, but could grab it later and do the fine detail, adding his personal touch to it. The scripter can also communicate with the community, get feedback, and keep tweaking the script as needed, something the AI model cannot do (yet) without human input.
Of course, maybe in a couple of years, we'd have a RealcumberAI trained with his scripts only. A corpus with the minor and personal tweaks he's made (and only his) could perfectly mimic his decisions around tiny twitches, movements, or anything happening in a specific frame. Of course, this opens the door to ownership, licensing, and many different things. But, yeah. It is inevitable.
