Prediction Promises: Towards a Metaphorology of Artificial Intelligence

Abstract

ABSTRACT Metaphors are a basic element of scientific language. In the context of Artificial Intelligence, metaphors facilitate the communication between experts and the public, as well as between scientists and engineers. Reimagining AI, here understood as a programmatic call, firstly requires us to be aware of the images we think with. Therefore, in this paper, I develop a claim for a hermeneutic metaphorology of AI, a philosophical investigation of our understanding of AI along metaphors. By studying metaphors of AI, we gain another tool for situating the discourse in and around AI towards its motivational background. This is important since our epistemic practices can’t be isolated from their political implications. Drawing mainly on Hans Blumenberg’s phenomenological philosophy of metaphor, this paper aims to mark the first step towards the broader metaphorological project. The focus hereby will lie on prediction as a paradigm of AI and a source for AI images. By analyzing two images, namely the expert and the black box, this article explores what these metaphors can tell us about the ways in which we perceive AI and how this affects the way we think about and practice human-machine relationality.