It's established science (see Chomsky) that a probablistic model will never achieve something approaching AGI