I could see a system where you very loosely define a handful of keyframes, give it some kind of description for what your looking for, and then the model fills in the rest for you. It almost certainly would still require some tweaking on the artists part, but I think a huge benefit is that AI can be non-deterministic. If you don’t like what you got you can just run it again without changing inputs. Maybe save multiple runs and manually combine all the best elements. Lots of opportunity.
A more complicated but ultimately faster approach is using a structure like an Entity Component System. You build an entity (deer, person, plant) out of components that are just data (health, hunger, mood), and then each type of component has a corresponding system that updates all the components at once based on other values. It’s somewhat similar, but you save space on unnecessary components not being added, and it packs the data together in way that is faster for the computer to iterate through.