Positive Surprise
An AI response that exceeds the user's prior expectations in quality, depth, or perspective. The referent is the discrete expectation-exceeding response event, which is associated with an increase in the user's trust in the collaboration.
Operational Definition
EVENT. One occurrence is recorded when a user rates a specific response as exceeding the expectation they held before receiving it. The observable marker is a positive expectation-disconfirmation judgement attached to an individual response (post-rating higher than the pre-stated or concurrently reported expectation).
Measurement Schema
Proposed measurement protocol (not yet empirically validated): Expectation-disconfirmation rating per response: users state an expectation level before a response and rate the delivered response on the same 5-point scale, yielding a signed disconfirmation score; positive surprises are responses with a positive score beyond a set margin. Could be reported as the rate of positive-surprise responses and mean disconfirmation; an optional paired trust item before and after gauges the associated trust shift (Wilcoxon signed-rank).