Pause Giant Anthropomorphizing Metaphors

Abstract The reemergence of pseudoscientific discourses around “AI” suggests using a terminology that combats retrograde biological essentialism. By conflating nodes within virtual software architectures, so-called convolutional neural networks, with biological neurons some computer scientists have fueled a surge of “AI” hype playing into the narrative that human-like artificial general intelligence (AGI) is near. The exaggeration of the technological capacity to produce human-level intelligence relies heavily on unsubstantiated metaphors that involve anthropomorphization. This discourse analysis builds on John Fiske's concept of technostruggles, whereby technoscientific discourse exercises power by producing “a particular form of social knowledge.” This article expands on how using anthropomorphizing metaphors enables pseudoscientific approaches such as phrenology and physiognomy. Consequentially the thinkpiece suggests a more self-reflective language around “AI” and de-anthropomorphized notions and concepts such as “weights” or “nodes” instead of “neurons.”