The AI Classification of Capability

Every society classifies, and the way it classifies determines what it can see, value, and sustain. Mauss and Durkheim's 1903 analysis of primitive classification showed that categorical systems are not discovered in the world but projected onto it from social organization; and that what falls outside the categories does not merely go unmeasured but ceases to be socially real—invisible, unrewarded, and therefore, gradually, unpracticed. Artificial intelligence is imposing a new taxonomy on the landscape of professional capability. The taxonomy divides human skill into two fundamental categories: the amplifiable—what the tool can process, accelerate, and scale; and the unrecognizable—what the tool cannot parse. The amplifiable is the category of embodied technique that translates into prompts, of tacit judgment that can be decomposed into explicit steps, of pattern recognition that can be specified in sufficient detail for the tool to reproduce its outputs. The unrecognizable is the category of what resists this translation: the feel for the codebase that operates below articulation, the diagnostician's sense