ABSTRACT
Are the concepts of “madness” and “normality” applicable to modern AI technologies? In this chapter, we will endeavor on a journey to explore how Foucault’s conception of “madness” along with Canguilhem’s distinction between the “normal and the pathological” can be utilized in an analysis of AI systems. Approaching AI from a philosophical, or even a psychoanalytic perspective, may seem prima facie bizarre or out of place. However, as we shall see AI applications may be considered as peculiar types of machines. Due to their proximity to the human, or shall we say the “man behind the machine,” we believe that some categories of thought which were reserved for organic life are quite relevant here. Canguilhem believed that there is no pathological machine. A machine can only malfunction. However, an organism at the pathological state is not an organism in a chaotic state, but in one with its own set of rules. What if such a set of rules emerges from a deep learning “machine”? Foucault was well aware of how normality is enforced and how madness is institutionalized. Using this as a metaphor, one could expand on how this institutionalization affects the opacity of AI, creating an obscure mechanical “madness.”
