Theorizing Mathematical Narrative through Machine Learning

Artificial intelligence (AI), machine learning (ML), neural networks— from smart technologies like Google Home or iPhone’s Siri, to predictions of batting performance in the MLB, to algorithmic bias prevention in hiring—are everywhere, threatening to displace the human.1 But what about these technologies is it that seems to so irreverently intrude upon our humanity? Kate Crawford claims that narratives about artificial intelligence, circulating widely for several centuries, have created and fortified the “myth [...] that nonhuman systems (be it computers or horses) are analogues for human minds” (4). Popular cultural artifacts, from Ridley Scott’s film Blade Runner to Nobel Prize-winning author Kazuo Ishiguro’s novel Klara and the Sun, continue Crawford’s historical narratives into the present day. The myth of the analogy of human brain and computer drives technical terminology as well: as computer scientist and information philosopher Brian Cantwell Smith explains, the architecture of the most recent wave of artificial forms of intelligence, embodied in machine learning techniques,2 is typically designated by the term “neural networks” “because of [its] topological similarity to the way the brain is organized at the neural level” (47). Thus, technical (or technical-seeming) definitions and mythologizing narratives converge in producing our dominant understanding of artificial intelligence, specifically of its machine learning components. Excitement about the promise of artificial intelligent systems as