How to Teach Machines in Artificial Intelligence? Technical Education in John Dewey, Gilbert Simondon and Machine Learning

Abstract

A society that uses drills and conditioning to train its machine learning models risks creating an alienated situation between human and machine in social life, as these teaching methods generate a lack of responsibility for the actions produced by such machines. Both John Dewey and Gilbert Simondon present conceptions and ideas that shed a different light on such alienated human–machine relationships in machine learning: that is, a critique of mere drilling and training as an educational method; the mode of existence of technical objects; and the importance of experience and practicality, for example, learning by doing. After giving an introduction to Dewey's and Simondon's conceptions, the analysis of case studies will show, among other things, that machine learning models need constant care by engineers and programmers. This allows us to highlight Simondon's and Dewey's critique of ancient philosophy. Whereas the latter separates thought and practice, labor and leisure, Simondon and Dewey foreground activities that are capable of bringing both together: Dewey's aesthetic experience and Simondon's technical activity.