Simone C. Niquille’s animation Homeschool allows the viewer to see through the eyes of a domestic floor cleaning robot as it experiences and learns to identify the environment around it. These domestic robots, which were developed in the late 1990s and have long entered our private homes, are programmed to move around chairs, along walls and on carpets. To be able to do so, they have to be trained via datatsets to learn what all the elements of a domestic environment are constituted of and how they can be identified. In this short narrative piece, Niquille playfully explores the mechanisms and processes employed to teach machines to make sense of visual inputs through machine learning algorithms, and questions the implications of categorising the world according to specific labels. Set within a virtual scenography assembled with the photorealistic images of one of the largest training datasets (SceneNet RGB-D), Homeschool demonstrates the ambiguity that emerges as a consequence of the friction between machine-generated models of reality and reality itself.
More by Simone C. Niquille: technofle.sh