Television and the Origins of Visual Pattern Recognition AI in Japan: a ‘Quest for a Seeing Machine’

What is it about?

This article examines the role of television in the history of contemporary data-driven computer vision. It does so through a focus on Japan's public broadcaster, NHK, in the 1960s. Specifically, the article examines NHK's Visual and Auditory Information Science Unit and its role in shaping the work of Fukushima Kunihiko, inventor of the world's first deep convolutional neural network. The use of television to harness data about viewer behaviour gave rise to the modelling eye-brain information processing, particularly mechanisms of feature extraction. Feature extraction in turn provided for Fukushima a link between his formative work on signal compression and the development of a pattern recognition machine. Recovering this history is important for two reasons. First, it helps counter a trend of “digital universalism” that homogenizes local differences into a single “algorithmic culture” of artificial intelligence in the Cold War United States. Second, it makes visible the largely ignored role of television within the genesis not only of digital image technologies, but information technologies more broadly.

Featured Image