CONCEPT
Double WYSIATI
The compounding blindness that emerges when two systems—each unable to flag its own gaps—collaborate to construct a story so coherent that neither participant can perceive what both are missing.
Daniel Kahneman’s acronym
WYSIATI—What You See Is All There Is—names System 1’s tendency to build the most coherent story it can from available information without experiencing the absent information as absent. The machine has its own structural equivalent: a
large language model constructs the best output it can from the patterns in its training data, with no mechanism for flagging where that data is thin, skewed, or simply wrong. When a human operating under WYSIATI evaluates output produced by a system operating under its own WYSIATI, the result is a double blindness whose combined confidence exceeds what either participant could generate alone. The human does not notice what the model omitted because the model’s output is too polished to suggest omission; the model did not notice the omission because it has no access to the context the human brings. Each party’s limitation conceals the other’s. The coherence of the product is higher than either could achieve separately, and the gaps are more deeply hidden than either