🤖 AI Summary
Existing generative models lack a well-defined fairness evaluation objective when demographic attributes are unspecified, leading to ambiguous assessment criteria. This work systematically formalizes this "missing target" problem and proposes a four-component framework for constructing evaluation targets—comprising the evaluation subject, prior acceptability, allocation mechanism, and operationalization—treating target distribution specification as an integral part of fairness evaluation rather than a preliminary step. Leveraging geographic and occupational priors, the study employs generative auditing and Jensen–Shannon divergence (JSD) to quantify bias, revealing significant deviations between model outputs and geographically derived targets (JSD = 0.508–0.606). Replacing these with uniform-probability targets alters per-cell JSD by 0.279–0.355 on average, underscoring the critical influence of target selection on fairness assessments.
📝 Abstract
Fairness evaluation concerns not only what a model produces, but also what its outputs ought to be compared against. When a model generates "a CEO in the United States," the prompt leaves demographic realization to the model. Existing group fairness definitions assume that sensitive attributes are given on the input side. Generative audits instead examine output-side demographic composition, yet the targets they compare it against are typically supplied rather than justified. The upstream question is what the target distribution should be. We formalize this missing-target problem for demographic-value-unspecified generation and decompose target construction into four commitments: the evaluative object, prior admissibility, allocation, and operationalization. In this framework, we admit the geographic prior under a geographic-membership interpretation for the declared public-world use. The occupational prior, under an incumbency interpretation, requires an independently defended objective such as workforce-composition fidelity. Instantiating this construction in AP-Bench, we find substantial distribution divergence from geography-derived targets, ranging from 0.508 to 0.606 on a 0-to-1 scale. Replacing each geography-derived target with an equal-category comparator, while holding generations and measurement fixed, produces model-specific mean absolute cell-level $\mathrm{JSD}_2$ changes ranging from 0.279 to 0.355. Target construction is therefore not a preliminary to fairness evaluation but a component of it. What we supply is not a universal target, but a framework that makes explicit the justification required before a distribution can serve as a fairness standard.