A Trustworthiness-Based Metaphysics of Artificial Intelligence Systems