Note: the image goes in and out of focus. Also, the audio is bad during Eric Owen Moss’s introduction, but improves at 5:05, as Manuel de Landa begins his talk.
Manuel de Landa contrasts the view that language gives form to the world – derived from Immanuel Kant – and the view that social practice gives rise to the world – derived from David Hume. The Humean tradition led to what De Landa describes as the old materialism, exemplified in the work of Karl Marx. De Landa proposes morphogenesis and emergent properties – discovered by chemistry, physics, biology and other hard sciences – are the basis of a new materialism, in which “it is impossible to view matter as an inert receptacle of forms that come from outside.” While the hard sciences have been exploring these concepts, De Landa argues that, in contrast, philosophy, cultural studies and the humanities in the 20th century were dominated by Kantian or Neo-Kantian thinking. He cites Heidegger, Husserl, Derrida, Kristeva, Baudrillard, and Foucault – excepting only Deleuze. This debate between Kant and Hume is relevant now because it underlies two approaches to Artificial Intelligence: Symbolic and Connectionist. De Landa argues that Connectionism creates a technological paradigm for “know how”, thinking without symbols. He characterizes neural networks as self-organizing dynamic systems.
The video ends before the end of the talk.