CONCEPT
Plausibility vs. Rightness
The critical distinction for AI evaluation—output can be plausible (conventionally correct, internally coherent) without being right (achieving fit, productivity, purpose-satisfaction).
Plausibility and rightness are not the same thing—this is the single most important diagnostic that Goodman's framework provides for the age of AI. Plausible output looks right: it is syntactically correct, structurally organized, conventionally appropriate, internally coherent. It passes surface tests. Rightness is deeper: it requires that the output achieve fit with the worldmaking project of which it is a part, produce genuine understanding rather than rehearsing familiar points, serve the purposes for which the version was constructed, and meet the substantive (not merely formal) standards of its symbol system. A plausible rendering can fail all these criteria while appearing indistinguishable from a right rendering to someone not evaluating with Goodman's multi-dimensional rigor. The danger AI poses to creative and intellectual work is not that it renders badly—it renders with spectacular plausibility. The danger is that the plausibility can substitute for rightness in environments where the evaluative attention required to distinguish them has atrophied, been outsourced, or simply overwhelmed by the volume of output demanding assessment.
