Critical appreciation is the cognitive stance Postrel's glamour framework prescribes for navigating AI: the ability to extract genuine value from the technology while remaining clear-eyed about what it conceals. It requires holding two truths simultaneously—AI genuinely expands capability (the functional outputs are real, the speed is real, the democratization of execution is real) and AI's presentation is glamorous (edited, idealized, concealing friction that returns). The stance is neither uncritical enthusiasm nor wholesale rejection but disciplined engagement: using tools with full awareness that polished outputs are not the same as polished thinking, that functional correctness is not architectural wisdom, that fluent prose can carry hollow arguments. Critical appreciation treats the AI demo as the fashion photograph—real dress, real model, real light, but strategically edited to conceal the team, the hours, the hundred discarded shots. The viewer who sees through the editing can appreciate the dress without believing the fantasy that she will effortlessly look like the model.
The stance develops through practice, not instruction. Builders who have used AI tools extensively develop it by encountering the gap between projection and reality: the prompt that seemed clear producing output that missed the intent, the generated code that was syntactically correct but architecturally wrong, the prose that sounded like insight but collapsed under scrutiny. Each encounter deposits a layer of caution—not cynicism, but calibrated trust. The experienced AI user knows which tasks the tool handles reliably and which require extensive human oversight. This knowledge is situational, tacit, and non-transferable—the same kind of knowledge that expert practitioners develop in any domain.
Critical appreciation is distinct from the three common responses to AI that Postrel's framework makes legible. The enthusiast sees only capability expansion and misses concealment—adopting tools with uncalibrated trust, experiencing disillusionment when outputs fail in ways demos never showed. The skeptic sees only concealment and misses genuine capability—rejecting tools that could expand her work because the glamorous presentation triggers justified suspicion. The critical appreciator sees both: genuine capability and systematic concealment, real expansion and relocated friction. The stance is cognitively demanding—holding contradictions without resolving them prematurely—but it is the only stance that allows full value extraction.
Organizations attempting to navigate AI adoption are, often without recognizing it, attempting to build critical appreciation at institutional scale. The Berkeley researchers' AI Practice framework (structured pauses, sequenced workflows, protected mentoring time) is implicit critical appreciation: benefit from the tool while maintaining the conditions that develop judgment the tool cannot provide. The mistake most organizations make is treating AI as pure capability expansion and discovering too late that capability without judgment floods them with output they cannot evaluate—the competency trap at AI speed.
The deepest application is to one's own outputs. The builder working with AI must become a critical appreciator of her own work—not assuming that because the code runs or the prose reads smoothly, the substance is sound. Segal's Deleuze fabrication story (Claude producing a philosophically wrong passage that sounded right) is the paradigm: smooth surface, hollow core, the gap visible only to sustained critical engagement. The builder with critical appreciation catches this before shipping. The builder without it ships confident inadequacy at scale.
The term is this simulation's coinage, synthesizing Postrel's glamour analysis with the practical challenge builders face. Postrel developed the glamour framework to explain fashion, politics, architecture. The simulation extends it to the specific challenge of evaluating AI outputs: tools that consistently produce glamorous presentations (polished, confident, conventionally correct) requiring the evaluative discipline of seeing through to substance.
The concept fills a gap in AI discourse. Enthusiasts say 'embrace the tools.' Critics say 'resist the seduction.' Neither stance is adequate. Critical appreciation is the third way: engage fully while maintaining the critical apparatus that distinguishes real value from projected ideal. It is the aesthetic version of Segal's beaver—building in the current rather than fighting it or being swept away.
Simultaneous seeing of capability and concealment. AI genuinely expands what humans can produce and systematically hides the friction, judgment, and labor required—both truths must be held without collapsing into enthusiasm or rejection.
Develops through encounter, not instruction. Critical appreciation is built by using tools extensively, experiencing the gap between projection and reality, calibrating trust through repeated evaluation of outputs.
Applied to one's own work most critically. The hardest application is evaluating outputs one produced with AI assistance—resisting the seduction of smooth surfaces that conceal substantive hollowness.
Organizational capacity, not just individual. Institutions must cultivate collective critical appreciation—structured practices ensuring the enthusiasm for speed does not eliminate the discipline of evaluation.