Since the initial development of transformer-based Large Language Models, we've been confronted with a fundamental philosophical question: do they really understand what they're saying, or are they blindly manipulating symbols?  This book argues that the answer to this question is . . . Yes.   Drawing on an inferentialist account of conceptual understanding, I argue for a radical break between sentience, understood as conscious awareness, and sapience, understood as conceptual understanding.  Non-linguistic animals like cats possess the former without possessing the latter.  We, of course, possess both.  LLMs, I argue, possess the latter without possessing the former.

To be published in (hopefully early) 2027 with Springer.  This book builds on my 2026 paper, "Sapience without Sentience"

Chapter Breakdown:

Chapter 1: The New Conversationalists

I propose a new version of the Turing Test, use it to define a Turing-based criterion of AGI, and preliminarily consider the implications of LLMs meeting it. pdf of chapter draft

Chapter 2: Sentience: An Enactivist Account

I develop an account of sentience, drawing on Aristotle, J.J. Gibson, and Heidegger.

Chapter 3: Sapience: An Inferentialist Account

I articulate an inferentialist account of sapience and consider some of its counter-intuitive consequences.

Chapter 4: Sapience without Sentience

I apply the previous accounts of sentience and sapience to current LLMs to argue that they are sapient without being sentient. 

Chapter 5: There Is No Symbol Grounding Problem.

I respond to objections from Searle and Harnad, arguing that knowledge of syntactic relations between symbols could in principle be sufficient for genuine semantic understanding.  Based on material given in this talk.

Chapter 6: Placing LLMs in the Space of Reasons

I argue that attributing sapience to LLMs taking a normative stance towards them---placing them in the "space of reasons"---and that is reasonable to take this stance towards them. Based on material given in this talk.

Chapter 7: Where Do We Go from Here?

I consider the ethical and safety implications of the view