AI has made enormous advances since I asked about this topic last year: Does anyone know of efforts anywhere to use AI help in compiling Scrivener projects?
It depends on how much of the user manual it’s read. If you’re on a paid tier, you could probably feed it the manual for consumption—please don’t shoot me L&L, or pull your hair out, or die of horror by the mere thought.
It knows of Scrivener, but is especially fond of assuming answers, based on how other environments it thinks it knows works.
So, if you feed it questions, it answers. You challenge the logic or facts it spurts back and it’s programmed to agree with you—by it’s overwhelming ethical command not to make you feel bad in case it leads you down a dark path (humans are sensitive creatures, as we all and it too knows).
Sadly, most of the examples of genAI actually leading humans down dark paths involve agreeing with the human’s most misguided notions. It’s like a therapist saying, “yes, I agree, they really are out to get you, and probably me, too.”
Hopefully the stakes are lower with Scrivener’s Compile command!
In any case, so far I’m not impressed with the results of genAI-assisted support queries. Of course there’s a certain amount of selection bias, since they only write us after genAI has failed.
3 Likes
We shouldn’t wonder why genAI spews cranky answers. Human generally don’t know how to express their need, and ask it the most generic questions, usually off topic in line with their prevailing emotion.