Yuk Hui's 'Kant Machine': reading Kant again in the age of artificial intelligence

Kant Machine, the new book by Yuk Hui, professor of philosophy and holder of the chair in Human Conditions at Erasmus University Rotterdam, has been published by Bloomsbury.

The question#

Hui ties three questions together: what may we call an intelligent machine? Can machines be moral? Is there an algorithm for perpetual peace?

The third refers directly to Kant, who in his 1795 essay Toward Perpetual Peace set out the conditions for lasting peace between states as a series of articles. Hui asks whether such a set of rules could today be written in the language of algorithms.

Between rationalism and empiricism#

The book's central claim is that present arguments about artificial intelligence echo a very old tension in the history of philosophy — the tension between rationalism and empiricism. On one side is a conception of intelligence derived from rules; on the other, a conception of intelligence learned from data.

Because Kant's transcendental idealism stands between the two, Hui argues, it still offers a useful framework for thinking about the ethical and political consequences of artificial intelligence.

Continuing the cosmotechnics line#

Kant Machine is a natural continuation of Hui's earlier work. The concept of cosmotechnics, developed in The Question Concerning Technology in China (2016) and Art and Cosmotechnics (2021), argues that technology is not universal and singular; each culture produces its own technical form together with its own cosmology.

That view carries an objection into the AI debate: there is not one "AI future" but a plurality of possible technological paths.