Images Make People. On the Power of Images in Artificial Intelligence

The discourse on artificial intelligence (AI) is a visually powerful one. This applies to both the concrete visual and the more conceptual artefacts. As unavoidable as the images orientating very different target groups are, the images are never safe from disruption, deception, illusion or magic. AI images are man-made images of socio-technical artefacts, but also man-made models for such artefacts. Last but not least, images provide reference-free simulations of a self-setting reality. An enlightened approach to the images of AI must make the effort to ‘spell out’ the images that tend towards rational abbreviation and to differentiate the levels of image-theoretical function associated with them (imitation, representation, simulation) in order to be able to discuss the content of statements. At the same time, a humanly self-critical clarification of the production conditions for the images of AI (genitivus subjectivus) is required, which must be embedded and linked as an element of a larger intellectual operation (recognizing and solving problems). In a first step, the article outlines the dimensions and functions of (human-made) images. In a second step, with recourse to Hans Blumenberg, these are expanded to include concepts such as world picture and world models, which provide an important interpretative framework for the images of AI (in genitivus objectivus—third step—and genitivus subjectivus—fourth step) and are presented and ‘read’ in excerpts. In the fifth step, the visual images are linked back to the linguistic images (metaphors) and the transfer function between the sources and targets of meaning in the AI discourse is critically reflected.