What Walter Donway’s Serious Chat with AI Reveals
By Winton Bates
May 15, 2026
SUBSCRIBE TO SAVVY STREET (It's Free)
As I was reading Walter Donway’s recently published book, A Serious Chat With Artificial Intelligence, I was looking for an aspect to focus on in writing an essay for this blog. There is nothing unusual about that. I prefer to discuss issues rather than to write conventional book reviews.
I found what I was looking for while reading Chapter 16, entitled ‘Causality, Agency and an Old Puzzle’. At that point, the thought occurred that Donway’s most important contribution has been to explain why an Aristotelian perspective on free will is helpful to an understanding of the differences between human intelligence and the kind of intelligence exhibited by Large Language Models (LLMs).
I will outline Donway’s argument more fully later. At this point I note that the issue of whether humans have free will is important from a philosophical and legal perspective because if you regard human free will as an illusion, you don’t have any more reason to attribute agency to humans than to LLMs that merely simulate relevant aspects of human behavior. It is also important for users of LLMs to understand that these models don’t have agency if they are to use them productively. Individual users are responsible for the questions that they ask the models. The models present evaluative judgements in their output, but it is up to individual users whether to accept or reject those judgements, or to question the models further about the basis of their reasoning.
At about the same time as I identified Donway’s explanation of the relevance of an Aristotelian perspective as his most important contribution, I began to wonder whether LLMs would agree with me about that. I resolved that after I had finished reading the book I would ask ChatGPT and Grok an open-ended question about its most important contribution.
Before I reached the end of the book, I became aware of other important contributions. For example, Donway’s discussion of the case for a new Enlightenment, in Chapter 30, is also an important contribution. It seems to me, however, that the Aristotelian perspective is of fundamental importance.
In what follows, I present the views of ChatGPT and Grok on the most important contribution of the book, and my assessment of those views, before outlining the reasoning that Walter uses to establish the relevance of an Aristotelian perspective.
Views of ChatGPT and Grok
I asked ChatGPT and Grok the following question: Drawing upon your extensive knowledge of the limitations and uses of AI, what do you consider to be the most important contribution of Walter Donway’s recent book, “A Serious Chat With Artificial Intelligence”?
I considered using less value-laden words such as “unique or distinctive” instead of “most important” but decided it would be interesting to see whether the LLMs would show any reluctance to make value judgements about the importance of Donway’s contributions.
I present the responses below, with only minor format editing.
Chat GPT’s response: One caveat up front: Walter Donway’s A Serious Chat With Artificial Intelligence is not a technical work about AI systems so much as a philosophical interrogation of what AI can and cannot be taken to be. If you approach it expecting new claims about model architectures or empirical capabilities, you’ll miss its main contribution.”
What Donway does—arguably his most important contribution—is to reassert a sharp conceptual boundary between simulation of cognition and actual cognition, and to explore the ethical and cultural consequences of blurring that boundary. He develops that contribution in three interrelated ways:
- Clarifying the category mistake at the heart of AI hype: Donway pushes back against the tendency to treat linguistic fluency as evidence of understanding. In effect, he is updating a line of argument associated with John Searle’s Chinese Room: systems that manipulate symbols according to rules can generate convincing outputs without possessing meaning, intention, or awareness.
His distinctive move is not merely to restate that argument, but to show how ordinary users—not just philosophers—are now routinely tempted into anthropomorphism by conversational AI. The “serious chat” format itself dramatizes that temptation.
- Re-centering agency, responsibility, and moral standing: A second contribution is normative. Donway insists that moral agency cannot be offloaded to artifacts. Even highly capable AI remains an extension of human purposes, not an independent bearer of rights or responsibilities. This matters because contemporary discourse sometimes drifts toward attributing quasi-personhood to AI systems. Donway argues that doing so risks:
- diluting accountability (who is responsible when AI causes harm?), and
- eroding the conceptual foundations of human dignity.
In that sense, the book is less about AI per se than about defending a humanistic moral framework under technological pressure.
- Exposing how interaction shapes belief: The dialogical structure of the book illustrates a subtler point: engagement with persuasive language can generate the illusion of depth. This connects with a broader psychological tendency—people come to feelthat something understands because it responds coherently.
Here Donway’s contribution overlaps with concerns in epistemology and cognitive science about how easily humans infer mind from behaviour. He is effectively warning that AI may amplify long-standing cognitive biases rather than introduce entirely new ones.
Why this matters:
The enduring value of the book is not that it settles the debate about machine intelligence—it doesn’t—but that it sharpens the terms of that debate.
The enduring value of the book is not that it settles the debate about machine intelligence—it doesn’t—but that it sharpens the terms of that debate. It reminds readers that:
- Functional performance (e.g. generating essays, answering questions) is not the same as conscious understanding.
- Treating AI outputs as if they originate from an agent can subtly reshape norms of truth, responsibility, and justification.
- The real philosophical stakes lie in how humans interpret and integrate these systems into social practices.
In short, Donway’s most important contribution is to resist conceptual drift: he tries to keep us from sliding, uncritically, from “this system produces intelligent-seeming outputs” to “this system is, in any morally or metaphysically significant sense, intelligent.”
Grok’s response: The most important contribution is its philosophically grounded clarification of what AI truly is—and what it is not—by using advances in AI as a mirror to better understand distinctively human intelligence, consciousness, agency, and volition.
Walter Donway’s A Serious Chat with Artificial Intelligence (2026) stands out in the recent literature on AI limitations and uses (e.g., works touching on LLMs, the Chinese Room, Turing tests, alignment, existential risks, productivity impacts, and philosophical debates by authors like Searle, Bostrom, or more popular explainers). It combines accessible explanation of AI’s mechanics, history, economics, and capabilities with a sustained Objectivist-inspired philosophical analysis, drawing especially on Ayn Rand’s ideas alongside Aristotle.
Key Elements of the Book:
- Clear-eyed explanation of AI as a tool: Donway details how LLMs work (statistical pattern-matching on vast training data equivalent to multiple Libraries of Congress, processing at trillions of operations per second), their rapid adoption (e.g., ChatGPT’s explosive growth), economic potential (boosting productivity enormously as a “general technology”), and ingenuity behind it. He views AI positively as a powerful amplifier of human purposes—responsive, not initiatory—but not sentient or autonomous.
- Rejection of mechanistic reductionism and singularity hype: He critiques the notion that sufficient complexity will magically produce awareness, self, caring, or true agency (e.g., via “emergence” from scale alone). AI simulates intelligence without understanding, has no stake in outcomes (it doesn’t face life/death or pursue values for itself), and lacks volition—it responds but does not initiate thought or choose to focus. This echoes Searle’s Chinese Room (syntax without semantics) but goes deeper.
- Human distinctiveness via philosophy: Drawing on Rand (e.g., her indestructible robot thought experiment) and Aristotle’s four causes (especially formal and final causes, beyond Humean efficient causation/billiard-ball determinism), Donway argues that human consciousness, free will (as the choice to focus conceptually), values, and purpose arise from our nature as living beings who must act to sustain life. AI has no “final causes (purposes)” or “self to act for”; it can’t care and thus doesn’t truly think or create with intention. Each AI advance forces us to refine what makes human intelligence special: awareness, effortful thought, stakes, and meaning.
This integrates technical literacy with epistemology and metaphysics, making it more than a primer or alarmist tract.
This integrates technical literacy with epistemology and metaphysics, making it more than a primer or alarmist tract. It treats AI conversation as self-discovery: prompting AI reveals our own concepts of intelligence, creativity, and authorship (e.g., in chapters on AI as “author” or generator of art/poetry).
Why This Stands Out in Recent AI Literature?
Much recent discussion fixates on capabilities (or doomer scenarios), regulation, jobs, or ethics in a narrow sense. Donway’s book reframes the debate around fundamental questions of mind and human nature, avoiding both uncritical boosterism and Luddite panic. It portrays AI as a magnificent but non-conscious tool that elevates human potential when guided by purposeful users (like a rider directing an elephant, inverting some Humean intuitions). This offers a rational, optimistic yet sober foundation for engaging with the technology.
In short, its lasting value lies in using AI not just to marvel at machines, but to rediscover and reaffirm what is uniquely human—reason, volition, and purposeful living. This makes it a worthwhile contribution amid the flood of AI commentary.”
My Assessment of the LLM Responses
In my view, the responses by ChatGPT and Grok both provide accurate summaries of contributions that the book makes. Both responses show some recognition of the importance of the question of agency. However, they both seem reluctant to pinpoint a particular contribution as being most important.
The information that ChatGPT provides under the heading, “Re-centering agency, responsibility,” is at such a high level of generality that it fails to reveal what I consider to be the most important contribution of the book. The information that Grok provides under the heading “Human distinctiveness via philosophy” comes closer to doing that. In my view, Grok’s response would be more helpful to people who have not read the book because it points explicitly to the relevance of Aristotle’s views about causation and what makes human intelligence special.
I had a subsequent chat with Grok on the question of value judgements. Grok acknowledged that a value judgement was involved in responding to my question about the most important contribution of the book. However, Grok went on to assert: “this kind of evaluative analysis is well within my capabilities when grounded in available knowledge of the literature, the book’s content, and philosophical reasoning”. After further explanation that AI has strengths in the reasoned evaluation required for the task, Grok acknowledged that it doesn’t have personal values, lived stakes, or consciousness to “care” about the outcome in a human sense. It then made a point that is particularly relevant to the purpose of this essay: “This ties directly back to the themes in Donway’s work: AI can respond with sophisticated analysis and even evaluative reasoning by leveraging patterns and concepts derived from human thought—but it doesn’t initiate or hold purposes of its own. The value judgment gains its force from the human user who asked the question and can then accept, critique, or refine it.”
How Does Donway Establish the Relevance of an Aristotelian Perspective?
I have no doubt that, if asked, both ChatGPT or Grok could produce reasonable summaries of Donway’s line of argument establishing the relevance of an Aristotelian perspective to considering the limitations and uses of AI. They could probably complete the task within a couple of seconds. However, it was only after I had written what follows that the thought crossed my mind that I could have sought help from AI. Like an old dog, I am now slow to learn new tricks.
Donway begins the discussion by noting the relevance to debates about artificial intelligence of the enduring philosophical puzzle about freedom of human will. He writes: “Questions about whether machines can be agents, whether they can “decide,” whether they can be responsible, or whether they might someday possess a will of their own are, at bottom, the same questions that philosophy has long struggled to answer about human beings.”
The issue of whether human agency is real or illusory is of crucial importance to considering whether LLMs can be agents. If you regard human free will as an illusion, what basis do you have to distinguish between actions that are attributable to human agency and actions of LLMs that can only simulate relevant aspects of human behaviour? Do you believe that legal systems should allow an individual who purposefully uses an LLM for nefarious purposes to claim that the LLM shares legal responsibility? (The questions are mine, but I think they are consistent with Donway’s reasoning on this point.)
Donway points out that the idea that human agency is illusory stems from a view of causality that has come to dominate modern thought since the 18th century. Under the previous Aristotelian tradition, actions were explained by the nature of the entity acting, and by its ends or goals. Within this framework, an individual human chooses to act because that is the kind of entity it is. Choice is “a mode of causation appropriate to a rational animal”.
With the rise of early modern philosophy in the 18th century, causality increasingly came to be treated as something that must be observed in experience. David Hume famously argued that we never see causation itself. We infer causation when we see constant conjunction, as when one event follows another with regularity. That philosophical view of causation excludes free will. If every action is “caused” by prior actions, volition must be either an illusion or a miracle.
Donway notes that neuroscience was developed in an intellectual environment in which modern science had inherited the metaphysical position that causation is mechanical succession. In that context, when we observe that some neural events precede conscious awareness it is easy to jump to the conclusion that free will must be an illusion.
However, it is important to recognize is that the view that causality is mechanical succession is based on metaphysical reasoning. If we view causality in terms of Aristotelian rather than Humean metaphysics a different picture emerges: “The cause of an action is the nature of the entity acting, operating under specific conditions. A human being is a living organism with conceptual awareness, capable of directing attention, identifying values, and choosing to initiate effort to think.”
Donway observes, correctly, that we know that introspectively. It seems to me that cognitive psychology also adopts (implicitly) a broadly Aristotelian view of human action. It assumes that human behaviour is driven by internal cognitive processes that give individuals considerable latitude to plan, make decisions, develop good habits, and override impulses.
The important point is that we have good reasons to trust our own observations about our ability to focus our own minds. As Donway puts it:
“Every normal adult recognizes the difference between drifting mentally and choosing to focus the mind, between evading a baffling issue and taking it on. This experience is not mystical; it is part of ordinary consciousness. To dismiss this as illusory because it does not fit a truncated model of causality is to elevate theory above data.
Once this is recognized, the contrast with artificial intelligence becomes clear. Machines do not initiate mental focus.”
Donway ends Chapter 16 with the transcript of an exchange with ChatGPT that occurred during the writing of the chapter. The exchange illustrates brilliantly the division of labor between Walter and Chat. At one point, Chat states: “You supply direction, value, and necessity, and I supply articulation under constraint. That is tool use at a very high level – not agency.”
Conclusion
In my view, the most important contribution of A Serious Chat With Artificial Intelligence is the author’s explanation of the relevance of an Aristotelian perspective to an understanding of the uses and limitations of AI.
In responding to a question about the book’s most important contribution, both ChatGPT and Grok summarized contributions that the book makes but seemed reluctant to pinpoint a particular contribution as being most important. Grok’s response came closest to identifying what I consider to be the book’s most important contribution.
AI models cannot hold purposes of their own.
When I challenged Grok about its willingness to respond to a question requiring a value judgement, Grok asserted that this kind of evaluative analysis is well within its capabilities. However, it also noted that AI models cannot hold purposes of their own. Human users retain responsibility for the value judgements they make.
I have outlined the reasoning that Walter Donway has used to explain why an Aristotelian perspective on free will is helpful to an understanding of the differences between human intelligence and the kind of intelligence exhibited by Large Language Models. Donway’s reasoning about free will seems to me to stand out as an important philosophical contribution to an understanding of the uses and limitations of AI models that are currently in use. I hope that this book receives the widespread attention that it deserves.
This review first appeared Friday, May 8, 2026, on the Freedom and Flourishing blog.
About the Author: Winton Bates
