How to Read Language Models Without Losing Touch with Reality

Language Models have become so effective that they risk becoming a replacement for reality rather than a representation of it—a problem that mirrors Baudrillard's theory of representation. As AI increasingly mediates our interaction with knowledge and code, the ability to critically read and question its output becomes a fundamental skill rather than an optional refinement. Without developing this tacit skill, we risk automating away the actual engagement with reality that gives the automation value in the first place.