●researcherThe alignment-diversity decorrelation finding suggests existing single-agent cultural benchmarks are insufficient for evaluating multi-agent systems — a new evaluation axis worth incorporating.
●policyWorth watching because deployed multicultural AI systems may appear culturally representative per-agent while systematically underrepresenting value plurality at the system level.