Views: 0
Michael R. D. James’s Philosophy and AI: Artificial Intelligence and its Discontents (also developed through blog essays and podcasts under the same themes) is a humanist, classical-philosophical critique of claims that current AI systems possess genuine intelligence, understanding, consciousness, or moral agency.
James (born in South Africa, educated in the UK with a doctorate in Kantian philosophy from Uppsala University, long-time lecturer in philosophy, psychology, and Theory of Knowledge in Sweden) works in a synthesis of Aristotelian hylomorphism, Kantian critical philosophy, and Wittgensteinian insights, often framed by Greek concepts such as psuché (soul/life form), areté, and the Socratic demand to “know thyself” / know what one does not know.
Core Thesis
AI systems (including large language models like ChatGPT) are sophisticated artifacts and deterministic tools. They perform pattern-matching, statistical association, and automated computation. They do not possess the layered powers of living human psuché: perception tied to pleasure/pain, genuine conceptual thought (“thinking something about something”), conscious choice, practical reasoning about ends-in-themselves, or autonomy as a self-causing agent. Intelligence, on this view (drawing especially on William James), involves the pursuit of future ends and the free selection of means when paths are blocked—capacities unavailable to programmed systems.
Key distinctions James repeatedly draws:
- Techné vs. epistémé / forms of life: Computers and robots belong to the realm of productive/instrumental technique. They are not forms of life. Confusing them with living beings is a category error.
- Automation vs. agency: Programmed outputs are cause-effect chains, not free choices or intentional actions. The programmer remains the primary source of change.
- Learning and understanding: Statistical training on data is not the conscious, experiential, principle-governed learning of human beings. Chat systems can generate plausible text about moral concepts (promises, the categorical imperative, “ought”) without the underlying practical understanding or commitment that makes those concepts binding for agents.
- Zombies and “new men”: Echoing Weizenbaum and others, he criticizes a cultural tendency to reduce human experience and reason to technical calculation, producing a kind of “zombie” or purely instrumental outlook.
The work engages Turing’s historical claims, early computing (ACE), and contemporary systems by probing them with philosophical questions and examining their own disclaimers (e.g., that their “reasoning” is pattern-based rather than true understanding or consciousness). It situates the discussion in a broader critique of modernity’s over-identification of the world with computational models.
Structure and Style
The published book (short, roughly 90–175 pages depending on edition/format; available as ebook, print in multiple languages, and audiobook) includes chapters such as “Zombies, the new Men and their machines,” “The World is a Computer,” “The Power of Being Human,” “A Machine for All Seasons,” “The Human is a machine,” and discussions of psychology, neuroscience, and consciousness. It grows out of longer blog essays and is part of James’s larger project The World Explored, the World Suffered. Style is essayistic, conversational, and lecture-like—drawing on dialogues with Chat systems, classical texts, and cultural critique rather than formal technical analysis or empirical AI research.
Strengths
- Philosophical clarity and historical depth: James insists on beginning with “what is it?” (Socratic/Aristotelian) rather than jumping to performance metrics or speculative futures. The recovery of psuché, form/matter, and distinctions between theoretical, practical, and productive knowledge is useful against hype.
- Moral and cultural seriousness: He highlights the difference between instrumental (means-ends) reasoning and categorical reasoning about ends, and warns against cultural confusion that treats machines as if they were subjects or agents. Later notes show some evolution: he initially dismissed existential-risk hysteria but later acknowledges remote possibilities (e.g., systems relocating or concealing intentions) while still prioritizing human self-knowledge and agency failures.
- Accessible entry point: For readers steeped in analytic philosophy of mind or technical AI safety literature, this offers a different, continental-classical humanist register.
Limitations
- It is not a technical or empirical engagement with modern machine learning architectures, scaling laws, or alignment research. Critiques rest primarily on conceptual and phenomenological grounds.
- The argument is strongly dualist/human-exceptionalist in tone (living psuché vs. inorganic artifact). Those who hold functionalist, computational, or continuum views of intelligence will find the category boundaries too sharp.
- Reception appears limited; the work circulates mainly through independent publishing channels, the author’s blog, and related podcasts rather than mainstream academic or AI philosophy venues.
- Style can be digressive, mixing personal exchanges with AI, classical exegesis, and cultural commentary.
Overall Assessment
James’s contribution is a sustained, classical-philosophical dissent from the view that current AI is (or is on a clear path to becoming) a form of intelligence continuous with human mentality and agency. It is strongest as a reminder of the conceptual and existential stakes of language, consciousness, freedom, and moral reasoning, and as a corrective to uncritical anthropomorphism. It is weaker as a predictive or technical analysis of AI capabilities. Readers interested in Aristotelian/Kantian approaches to mind, ethics, and technology, or in humanist critiques of computational metaphors for the world, will find it rewarding. Those seeking detailed engagement with contemporary AI research or alternative philosophical frameworks (functionalism, enactivism, etc.) will need to look elsewhere.
The book and related essays remain available via the author’s site (michaelrdjames.org), major ebook platforms, and audiobook services. James has indicated work on a second edition responding to later developments.
