is it cynical that i find the idea of a "latent space" a bit goofy? may just be because i think data science is kind of a joke and much machine learning is "we threw shit at our model but don't know what a category error is"

like it's just dimensionality reduction with new vocab?