Clever But Not Wise: The Limits of AI and the Space of the Human
TL;DR
It is argued that while Generative AI is clever, it is not wise, and that a world governed by AI would be a world that had gained in cleverness, but had lost wisdom.
Abstract
Abstract This paper argues that while Generative AI is clever, it is not wise. Cleverness involves a capacity for instrumental rationality – a technocratic capacity to generate means towards externally specified ends – but wisdom involves a capacity for reflectively deliberating about ends, weighing them up against each other, and acting on them even in the absence of generalised decision procedures. The former is the province of AI reasoning, while the latter is the province of human reasoning. Drawing on Stuart Russell’s distinction between binary and unary functions – where it is argued that AI is limited to a binary function that involves creating means to fulfil an exogenously specified end – this paper claims that we can draw an operationalised distinction between these two forms of reasoning that would show up in observable differences between how human agents and AI systems act. The paper concludes that a world governed by AI would be a world that had gained in cleverness, but had lost wisdom.