In their open-access publication, Making AI Intelligible: Philosophical Foundations (Oxford University Press, 2021), Herman Cappelen and Josh Dever argue that philosophers of language can contribute to a deeper understanding of artificial intelligence. AIs known as “neural nets” are becoming commonplace and we increasingly rely on their outputs for action-guidance, as when an AI like Siri hears your question and says, “There’s a pizza shop on the corner.” Our use of words like “says” suggests an important question: do AIs literally say anything? Should we understand their outputs as utterances with meaningful content? And if so, what makes that content meaningful, and how is it related to the processes which result in that output? Cappelen and Dever take up these questions and propose a framework for answering them, abstracting from existing externalist approaches to develop a “de-anthropocentrized” externalism for AI. The book introduces readers not only to issues in AI surrounding its content and interpretation, but also to concepts in philosophy of language which may be relevant to these issues, serving as an invitation for further investigation by philosophers and programmers alike.

Malcolm Keating is Assistant Professor of Philosophy at Yale-NUS College. His research focuses on Sanskrit philosophy of language and epistemology. He is the author of Language, Meaning, and Use in Indian Philosophy (Bloomsbury Press, 2019) and host of the podcast Sutras (and stuff).