Helping Machines (Help Us) Make Mistakes

narrativity in generative art | full text


The development of automatic narrative systems has been largely driven by the engineering tendency to anthropomorphize the machine logic so they can ‘tell stories’ similar to how humans do. From the artists’ perspective, however, the experimentation with their media is often more important than the (plausibility of) storytelling, and it often unfolds in non-verbal events that have a potential to generate diverse narratives through experience of the audience. We discuss the emergence of the creative practices that enrich the poetic repertoire of new media art by playfully utilizing the machine flaws, irregularities, errors and systemic technical imperfections thus revealing the human biases and fallacies entangled with technology. One of the implications of these practices is that if the AI research opens up a broader space in which a machine could achieve its own authorial voice, our concept and understanding of the narrative would need to be reconsidered.

Keywords: Algorithm, Artificial Intelligence, Error, Generative Art, Machine Learning, Narrativity, New Media Art, Programming, Software.

Publication (with Vladimir Todorović): xCoAx 2019 conference proceedings. Milan, 2019.