When algorithms assess and allocate work to people and AI systems, machines become judges in a Reverse Turing Test. We argue that suitability assessments should compare profiles of underlying capabilities and propensities, rather than rely on incompatible measures such as human qualifications and AI benchmark scores. The paper outlines how a shared approach can support predictions about new tasks, explain differences in performance, and make assessments easier to audit.
This is a position-paper preprint archived in May 2026.