So from what i understand one of the main problems with training an effective AI is needing a gigantic training dataset. Maybe we could enlist the communities help with that, folks would upload their highpoly and retopologized mesh which we could use to train the AI. As you can probably tell i absolutely hate the tedious process of retopologizing. And i know i am not the only one, maybe people who hate it just as much as i do will contribute their past projects for the cause?
The first step would be to research an algorithm that can actually do good retopology using AI, which as far as I know does not exist.
AI is not something you can just apply to any problem.
Well deep learning is a fairly general purpose algorithm that has been applied to many types of problems before with mixed results.
The problem is that you need a big dataset to find out whether an algorithm is good enough or not for some type of problem in the first place. So without trying it’s impossible to know if it will work.