Talk with James Turner (Umeå University)
Functions and Representations in AI Systems
16–17 November 2026
Carlsberg Academy, Copenhagen, Denmark
Organized by the Center for Philosophy of Artificial Intelligence, University of Copenhagen, with support from the Pioneer Centre for Artificial Intelligence.
Abstract
Can artificial neural networks have proper functions, and can these functions ground genuine internal representations? We approach these questions through the familiar teleosemantic example of frogs detecting and responding to prey, considering how analogous representational structures might arise in artificial systems.
We examine how historical processes of training, selection, and retention can ground proper functions in neural networks and how these functions relate to the individuation of internal representations. In particular, we investigate the extent to which existing etiological theories can accommodate artificial systems whose functional organization is shaped by optimization rather than biological reproduction.
The discussion bears on the relationship between biological and artificial representation, the explanatory role of proper functions, and the applicability of teleosemantics to contemporary AI systems.
