> Could still be a while before a breakthrough comes from any direction.

It would be valuable enough that getting significant funding to work on it is probably possible. Especially with all the money being thrown at AI.

Could you elaborate on what you mean by OOD detection seeming ill-posed?