What do you think?
Kant Machine: Critical Philosophy after AI
Rethinking the philosophy of Immanuel Kant in the age of artificial intelligence.
What could be called an intelligent machine? Are machines capable of being moral? Does an algorithm for perpetual peace exist? In this groundbreaking new work, Yuk Hui considers how current debates on artificial intelligence echo historical philosophical discussions about the workings of the mind, with the philosophy of Immanuel Kant emerging as a lens through which to consider the ethical and political implications of AI and robotics in a new light.
Addressing fundamental questions around machine intelligence and morality, transcendental idealism and learning, and the metaphysics of machines, the history of AI and Kantian ideas are expertly woven together alongside an array of figures in the histories of technology and from Gottfried Wilhelm Leibniz and Alan Turing to Hubert Dreyfus and Jacques de Vaucanson.
In asking how we can understand AI in light of the challenges Kant posed to both rationalism and empiricism, and how revisiting Kant can help us better comprehend the nature and limitations of contemporary technologies, Kant Machine is an essential critical contribution both to Kant studies and to the philosophy of digital technology.
What could be called an intelligent machine? Are machines capable of being moral? Does an algorithm for perpetual peace exist? In this groundbreaking new work, Yuk Hui considers how current debates on artificial intelligence echo historical philosophical discussions about the workings of the mind, with the philosophy of Immanuel Kant emerging as a lens through which to consider the ethical and political implications of AI and robotics in a new light.
Addressing fundamental questions around machine intelligence and morality, transcendental idealism and learning, and the metaphysics of machines, the history of AI and Kantian ideas are expertly woven together alongside an array of figures in the histories of technology and from Gottfried Wilhelm Leibniz and Alan Turing to Hubert Dreyfus and Jacques de Vaucanson.
In asking how we can understand AI in light of the challenges Kant posed to both rationalism and empiricism, and how revisiting Kant can help us better comprehend the nature and limitations of contemporary technologies, Kant Machine is an essential critical contribution both to Kant studies and to the philosophy of digital technology.
142 pages, Hardcover
Published February 5, 2026
About the author
Yuk Hui
29 books170 followersYuk Hui studied Computer Engineering and Philosophy at the University of Hong Kong and Goldsmiths College in London, with a focus on philosophy of technology. He currently teaches at the Bauhaus University in Weimar. Between 2012 and 2018 he taught at the institute of philosophy and art (IPK) and Institute of Culture and Aesthetics of Digital Media of the Leuphana University Lüneburg where he wrote his habilitation thesis. He is also a visiting professor at the China Academy of Art where he teaches a master class with Bernard Stiegler every spring. Since 2019 he is Visiting Associate Professor at the School of Creative Media of City University in Hong Kong. Previous to that, he was a research associate at the Institute for Culture and Aesthetics of Media (ICAM), postdoctoral researcher at the Institute for Research and Innovation of the Centre Pompidou in Paris and a visiting scientist at the Deutsche Telekom Laboratories in Berlin. He is initiator of the Research Network for Philosophy and Technology, an international network which facilitates researches and collaborations on philosophy and technology. Hui has published on philosophy of technology and media in periodicals such as Research in Phenomenology, Metaphilosophy, Parrhesia, Angelaki, Theory Culture and Society, Cahiers Simondon, Deleuze Studies, Intellectica, Krisis, Implications Philosophiques, Jahrbuch Technikphilosophie, Techné, Zeitschrift für Medienwissenschaft, Appareil, New Formations,Parallax, etc. He is editor (with Andreas Broeckmann) of 30 Years after Les Immatériaux: Art, Science and Theory (2015), and author of On the Existence of Digital Objects (prefaced by Bernard Stiegler, University of Minnesota Press, March 2016), The Question Concerning Technology in China. An Essay in Cosmotechnics (Urbanomic, December 2016), and Recursivity and Contingency (Rowman & Littlefield International, February 2019). His writings have been translated into a dozen languages.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 3 of 3 reviews
January 19, 2026
Such a disappointment. It is not a Kantian AI book in the sense that is structures AI through a critical lens.
I except a deeper and clearer exploration of the ontological structures of machines, since the author explicitly describes the three Kantian question within the introduction. After that point, up to which I was excited to continue reading, the author drops the ball by going into a deep monological description of the three Kantian Critiques.
My question is: Who is this book for? For philosophy post-docs or for the AI engineer? If it is targeted at the former, then he should give us a few guidelines on the Kantian conception of freedom (which he sees as self evident as has been described in the Critique of Practical Reason) and then analyse AI. If it is for the latter, then it needs to be toned down A LOT.
The language use itself is messy and the constant use of citations gives me the impression that Hui is unsure in his conclusions. This is connected to the work of the editors. I’d like to see the pre-edited version if they allowed this mess to be published!
The worst part was the equalization of AI harms and successes by saying that its functions are“pharmacological” (p.78). The current negative impacts are just be accepted on a utilitarian scale. He brings forth a fatalism that is unimaginable!
One could have summarized content to 10 pages with adequate footnotes. Seems more like the work of an undergrad who is desperate to fill words and present their knowledge of the Kantian works. Hui attempts to neither be a sceptic nor a dogmatist but remains as a bland nothing.
If our goal now should be to look at critical philosophy again - which the author wants us to do - then I don’t know why I’ve read the work at all. Wasn’t that the point of this book?
My day is ruined!
I except a deeper and clearer exploration of the ontological structures of machines, since the author explicitly describes the three Kantian question within the introduction. After that point, up to which I was excited to continue reading, the author drops the ball by going into a deep monological description of the three Kantian Critiques.
My question is: Who is this book for? For philosophy post-docs or for the AI engineer? If it is targeted at the former, then he should give us a few guidelines on the Kantian conception of freedom (which he sees as self evident as has been described in the Critique of Practical Reason) and then analyse AI. If it is for the latter, then it needs to be toned down A LOT.
The language use itself is messy and the constant use of citations gives me the impression that Hui is unsure in his conclusions. This is connected to the work of the editors. I’d like to see the pre-edited version if they allowed this mess to be published!
The worst part was the equalization of AI harms and successes by saying that its functions are“pharmacological” (p.78). The current negative impacts are just be accepted on a utilitarian scale. He brings forth a fatalism that is unimaginable!
One could have summarized content to 10 pages with adequate footnotes. Seems more like the work of an undergrad who is desperate to fill words and present their knowledge of the Kantian works. Hui attempts to neither be a sceptic nor a dogmatist but remains as a bland nothing.
If our goal now should be to look at critical philosophy again - which the author wants us to do - then I don’t know why I’ve read the work at all. Wasn’t that the point of this book?
My day is ruined!
September 13, 2026
The intersection of Kant and AI is fertile ground, and this is the second book I read on this interesting intersection (after Yapay Zekayı Kant’la Düşünmek (in Turkish)).
For me it was an interesting reading experience, because even though I haven't read Kant's third critique yet (Critique of the Power of Judgment) I've read the first two and have been studying them for a while. On top of that, having an academic background and strong personal interest in mathematics, logic, computer science and cognitive science certainly shaped my perception and the criticism that follows.
I can clearly say that not only some parts of the book seems to be rushed (incentives? editors?), but also some abuse of terminology started to make me nervous. How come? Well, if you had your share of computer science and programming, you will hopefully have a good understating of what "recursive" and "recursion" means (and if you didn't, then I suggest the following article: "How Not to Teach Recursion").) I'd expect more rigorous use of terms, instead of hand waving that almost feels like "recursive self improvement, therefore, magic, qualitatively different, etc.", which is more worth of Silicon Valley tech bros (you get what I mean if you spend more than a few weeks on That Part of Twitter (stupidly abbreviated as TPOT)).
To put it shortly, no matter how much I'm impressed by the recent mind blowing theoretical and technological advances of modern, state-of-the-art conversational AI systems, I find the author's argumentation regarding the qualitative, in-principal/categorical difference forced. And no invoking Gödel is not the move you think it is. (One more thing: self-modifying AI systems aren't something new, e.g., we had genetic programming for a long time with mind blowing applications. Nobody dared assign some deeply qualitative difference to these fantastic but in a sense boring systems.)
For the second part of the book, where the author takes Kant's morality and free will critique and explores them within the perspective modern AI, the argumentation felt a bit more persuasive but then again, the problem of "agency" is the Gordian Knot and unfortunately the word "agentic" once again, gained a very weird meaning in AI circles, even leaking into philosophical discussions, creating a mess. Having said that, even though I have my reservations, I agree with author to a large extent, in the sense that "moral machines" is not something we have and even the question/phrase itself seems ill-posed.
As for the final and third chapter, in which the author takes Kant's ideas about world governance and world peace, I don't know what to say. It read like a hodgepodge of ideas, failing to establish a strong narrative. Moreover, not having a single version of game theory, iterated prisoner's dilemma, etc. seemed very strange to me. I mean, if you'll explore this topic purely philosophically, that's OK, but if you'll bring computationalism to the table, ignoring game theory is really strange.
One more thing: in one of the footnotes, the author stressed the difference between "uncomputable" and "incalculable". I believe he could've gone further by talking about the question of "formalization", e.g., why there is such a strong belief that "formalization" is the the be-all and end-all of human thinking? I mean, yeah, "The Unreasonable Effectiveness of Mathematics in the Natural Sciences" and all that, but come on, we're doing philosophy, right? ;)
I'm glad I read that book, because some parts of it annoyed me so much that I will continue exploring the intersection of Kant and AI in more detail, and this, I consider a win.
For the curious readers, here's some further reading that I noted for myself:
- Are Emergent Abilities of Large Language Models a Mirage?
- Unexplainability of Artificial Intelligence Judgments in Kant’s Perspective
- Why Heideggerian AI failed and how fixing it would require making it more Heideggerian
- Circling the void: Using Heidegger and Lacan to think about large language models
For me it was an interesting reading experience, because even though I haven't read Kant's third critique yet (Critique of the Power of Judgment) I've read the first two and have been studying them for a while. On top of that, having an academic background and strong personal interest in mathematics, logic, computer science and cognitive science certainly shaped my perception and the criticism that follows.
I can clearly say that not only some parts of the book seems to be rushed (incentives? editors?), but also some abuse of terminology started to make me nervous. How come? Well, if you had your share of computer science and programming, you will hopefully have a good understating of what "recursive" and "recursion" means (and if you didn't, then I suggest the following article: "How Not to Teach Recursion").) I'd expect more rigorous use of terms, instead of hand waving that almost feels like "recursive self improvement, therefore, magic, qualitatively different, etc.", which is more worth of Silicon Valley tech bros (you get what I mean if you spend more than a few weeks on That Part of Twitter (stupidly abbreviated as TPOT)).
To put it shortly, no matter how much I'm impressed by the recent mind blowing theoretical and technological advances of modern, state-of-the-art conversational AI systems, I find the author's argumentation regarding the qualitative, in-principal/categorical difference forced. And no invoking Gödel is not the move you think it is. (One more thing: self-modifying AI systems aren't something new, e.g., we had genetic programming for a long time with mind blowing applications. Nobody dared assign some deeply qualitative difference to these fantastic but in a sense boring systems.)
For the second part of the book, where the author takes Kant's morality and free will critique and explores them within the perspective modern AI, the argumentation felt a bit more persuasive but then again, the problem of "agency" is the Gordian Knot and unfortunately the word "agentic" once again, gained a very weird meaning in AI circles, even leaking into philosophical discussions, creating a mess. Having said that, even though I have my reservations, I agree with author to a large extent, in the sense that "moral machines" is not something we have and even the question/phrase itself seems ill-posed.
As for the final and third chapter, in which the author takes Kant's ideas about world governance and world peace, I don't know what to say. It read like a hodgepodge of ideas, failing to establish a strong narrative. Moreover, not having a single version of game theory, iterated prisoner's dilemma, etc. seemed very strange to me. I mean, if you'll explore this topic purely philosophically, that's OK, but if you'll bring computationalism to the table, ignoring game theory is really strange.
One more thing: in one of the footnotes, the author stressed the difference between "uncomputable" and "incalculable". I believe he could've gone further by talking about the question of "formalization", e.g., why there is such a strong belief that "formalization" is the the be-all and end-all of human thinking? I mean, yeah, "The Unreasonable Effectiveness of Mathematics in the Natural Sciences" and all that, but come on, we're doing philosophy, right? ;)
I'm glad I read that book, because some parts of it annoyed me so much that I will continue exploring the intersection of Kant and AI in more detail, and this, I consider a win.
For the curious readers, here's some further reading that I noted for myself:
- Are Emergent Abilities of Large Language Models a Mirage?
- Unexplainability of Artificial Intelligence Judgments in Kant’s Perspective
- Why Heideggerian AI failed and how fixing it would require making it more Heideggerian
- Circling the void: Using Heidegger and Lacan to think about large language models
March 17, 2026
Ideas curiosas, pero se podría llevar más allá. La conclusión es que la pregunta sobre si las máquinas (y las IAs) pueden ser críticas queda abierta, como admite el propio Hui. Entonces, ¿para qué tanto rodeo?
Es un libro con tesis interesantes, aunque parece hecho con prisas a ratos. Hay saltos bruscos entre temáticas y digresiones que solo por casualidad conducen a formulaciones fecundas. Lo más molesto para mí es esa tendencia de Hui a incurrir en eternos desarrollos doxográficos para exponer ideas que se podían exponer sin acudir a la historia de la filosofía en absoluto.
Valoro positivamente los esfuerzos que está llevando a cabo para vincular tecnología y filosofía, pero creo que tiene que cambiar la estrategia a la hora de escribir. Al final en este tipo de trabajos el interesado en la tecnología solo solo acaba percibiendo divagaciones innecesarias y el interesado en la filosofia solo acaba percibiendo tergiversaciones forzadas.
Con todo, es una lectura estimulante que revisa algunos puntos no aclarados del todo bien en Recursividad y contingencia. El mayor pecado de este libro es haber sido producido en medio de un proyecto filosófico que todavía se encuentra en construcción.
Es un libro con tesis interesantes, aunque parece hecho con prisas a ratos. Hay saltos bruscos entre temáticas y digresiones que solo por casualidad conducen a formulaciones fecundas. Lo más molesto para mí es esa tendencia de Hui a incurrir en eternos desarrollos doxográficos para exponer ideas que se podían exponer sin acudir a la historia de la filosofía en absoluto.
Valoro positivamente los esfuerzos que está llevando a cabo para vincular tecnología y filosofía, pero creo que tiene que cambiar la estrategia a la hora de escribir. Al final en este tipo de trabajos el interesado en la tecnología solo solo acaba percibiendo divagaciones innecesarias y el interesado en la filosofia solo acaba percibiendo tergiversaciones forzadas.
Con todo, es una lectura estimulante que revisa algunos puntos no aclarados del todo bien en Recursividad y contingencia. El mayor pecado de este libro es haber sido producido en medio de un proyecto filosófico que todavía se encuentra en construcción.
Displaying 1 - 3 of 3 reviews