This work is a conversation between Rachel Smith (human) and cifar10_cnn.py (artificial neural network). In order to communicate successfully, they must speak the same language. The method of communication is a human/machine compromise; a hand-painted grid of pixels. The human must automate and restrict herself, the network must cope with human error and bodily forms – colors mixed with naked eyes and brushstrokes. By engaging with a closed system, Rachel forces a dialogue where there would usually be none. In exposing the “hidden layers” of the neural network, extracting digital information and replacing it with an analogue counterpart, she is able to influence the final decision made by the machine.
Biography:
Rachel Smith is a resident of the Pervasive Media Studio in Bristol, UK and is currently a research fellow in Automation for the South West Creative Technologies Network (SWCTN). She recently graduated from the Bauhaus University in Weimar with a master’s in media art and holds a bachelor’s degree in mathematics.
