Deep Learning as a black box, “OpenAI” and the Internet. On the media-political dimension of machine learning
Internet-based digital culture is increasingly shaped by modern technologies of artificial intelligence, with especially approaches of Deep Learning, which follow the functioning of biological brains, being in great demand today. As methods of machine learning, these approaches enable a computer to learn without being explicitly programmed for this purpose. This research project analyzes the media-political dimension of machine learning, and focuses on two aspects in particular. First, the question of how far Deep Learning represents a black-box technology, i.e. can be regarded as a system whose internal operations and functions are closed to human understanding. Second, the investigation of industrial-scientific projects (e.g. the non-profit research company OpenAI) that, at least according to their own claims, strive to democratize artificial intelligence. The project not only examines whether the claims are being met, but also analyzes the different media conditions of machine learning and locates them theoretically, historically and politically.
