Is it me or do dependent types look really similar to the axiom of choice?