Looking Outwards 06: Randomness

Robbie Barrat’s Neural Network Balenciaga series is a fascinating
amalgamation of AI, fashion, and the fine line between creativity
and mimicry. Barrat utilized new neural network processing methods
to analyze the similarities in thousands of images from Balenciaga
runway shows and lookbooks and pushed this data to another one of
his neural networks, prompting the system to interpret what makes
Balenciaga Balenciaga. The results are random but creative
mishmashes of what the AI learned from thousands of images and
is trained to think Balenciaga looks like.

I was especially drawn to this work because of how it throws the whole
concept of randomness into question – the AI may generate random iterations
of Balenciaga outfits, but the iterations look so similar to each other
that it makes the viewer ponder about the original source material itself.
I additionally was interested in this glimpse into how generative algorithms
really work – the AI doesn’t know what Balenciaga or fashion really is, yet
tries to replicate these foreign concepts to its best approximation and
succeeds quite well.