ABSTRACT
Let us call a human-to-machine interactive perception-based interagency attributive language-learning and -using system of the kind sketched here a system with a ‘PAĽ (Pure Agency attributive Language-learner). Now if a PAL is buildable, as it would seem to be, would it still have a frame problem? Would it still lack common sense? Would it still lack tacit knowledge? Would there still be a central processing vacuum? And, would its modules be essentially unconnected and unintegrated?
