Wrocław University of Science and Technology, Poland This article is a review of the book Making AI Intelligible. Philosophical Foundations, written by Herman Cappelen and Josh Dever, and published in 2021 by Oxford University Press. The authors of the reviewed book address the difficult issue of interpreting the results provided by AI systems and the links between human-specific content handling and the internal mechanisms of these systems. Considering the potential usefulness of various frameworks developed in philosophy to solve the problem, they conduct a thorough analysis of a wide spectrum of them, from the use of Saul Kripke’s work to a critical analysis of the explainable AI current.
According to the standard interpretation of Lewis’s theory of predicate meaning (the U&N theory), the naturalness of meaning candidates should be stated metaphysically as a length of definition in terms of fundamental properties. Recently, Weatherson has criticized the U&N theory and argued that the criterion of naturalness should be stated epistemologically as the amount of evidence needed to form a belief. Despite the criticism, his attitude towards the U&N theory is quite relaxed. According to Weatherson, the U&N theory can be used as a good heuristic for delivering the correct verdicts when doing applied semantics, i.e., when we try to determine the best meaning candidate for a particular predicate. In this paper, I try to show that the “good heuristic strategy” is of no use because A) there is no guarantee that the epistemological and the metaphysical criteria of naturalness deliver the same verdicts and B) even if they deliver the same verdicts, the difference in their theoretical backgrounds may affect arguments which rely on the verdicts. The difference will be shown by drawing on the example of Theodore Sider and his use of the U&N theory.
JavaScript is turned off in your web browser. Turn it on to take full advantage of this site, then refresh the page.