Helena Nikonole

»deus X mchn« explores issues around online security within the Internet of Things and the increasing capabilities of Artificial Intelligence. The piece is constituted of an IP (Internet Protocol) printer that automatically prints a text every 30 minutes. These texts, which at a first glance may not seem very different from famous sacred texts, are actually written by a machine: a neural network constituted of a long short-term memory (LSTM) algorithm, has been trained in the language of a corpus of sacred texts such as the Bible, Quran, Torah, Ramayana, Tao Te Ching, and others. The neural network perceives these texts as a sequence of numbers and performs computations in order to discover their grammatical structures; that is, the »code« of the language. By learning these semantic structures, it starts to generate its own text, trying to keep the essence of the religious documents.

Hacked IP cameras with speakers in public spaces anywhere in the world start reciting these texts to unsuspecting people, whose reactions are projected on the screen. The voice that reads out the texts is a second neural network that synthesizes a human voice that is embodied in an unsecured device connected to the Internet of Things.