What can we learn from artificial intelligence about video art? Research on the Vasulka Live Archive

Abstract

The paper introduces the conceptual and methodological background, as well as the results, of applying a specially trained AI tool for video art analysis, available at Vasulka Live Archive.net (Horáková, Schimmel, Sikora, Miklánek & Barok, 2021). It does not delve into the technical aspects of the intelligent software. Instead, it focuses on: a) the conceptual framework and methodology in which the project is situated (digital humanities), to uncover and justify the decisions that led to the design of the AI tool; and b) provides a report on the AI tool’s application in a service of data-based verification of generally accepted video art definitions by Peter Weibel and Rosalind Krauss. A digital archive of the Vasulkas’ videos, considered a classic example of video art, served as the research sample in this experiment. The application of the AI tool in a service of video art analysis is based on the assumption that if the AI tool is thought to be able to identify certain formal qualities and the leitmotifs across a video art archive, then it can be used for experimental, quantitative verification of theoretical statements about video as well. Thus, the results of applying the AI tool described in the paper provide a unique, quantitative meta-analysis of video art discourse.