You On AI Field Guide · Rigid Designation The You On AI Field Guide Home
TxtLowMedHigh
CONCEPT

Rigid Designation

Kripke’s term for the property of a name that fixes its object across all possible worlds—the device that lets thought hold an individual constant while the world varies, and whose structural absence from context-sensitive embeddings predicts the entity-tracking and coreference failures of current language models.
A rigid designator, Kripke wrote in Naming and Necessity, is “something that in every possible world designates the same object.” The point of rigidity is to secure sameness of subject across change of circumstance: when we ask what would have happened if Nixon had done otherwise, we need the name “Nixon” to keep gripping that same man through every counterfactual variation we consider, so that we are still talking about him and not drifting to whoever else might have occupied his circumstances. Proper names are rigid designators; definite descriptions are not—“the President of the United States in 1970” picks out a different person in a world where someone else won the election. This asymmetry reorganizes the entire philosophy of reference, overturning the Frege-Russell-Searle view that a name abbreviates a bundle of descriptions in the speaker’s head. And it lands with unusual force on the architecture of large language models, whose
← Home0%
CONCEPTBook →

Keep reading with YOU ON AI

Unlock the full book, 10,000+ field-guide entries, and a 1000+ thinker library. If you have a book code, register now — it takes a minute.

Register with book code Sign in