A Diagnostic Framework for Socially Sustainable AI Diffusion

Artificial intelligence (AI) promises large productivity gains, yet growing concern surrounds its implications for social sustainability. This paper develops and empirically evaluates a simple behavioral framework in which unequal access to AI generates mutually reinforcing gaps in economic performance and social visibility, potentially undermining the long-run stability of social systems. Individuals fall into two groups—AI adopters and non-adopters—and differences in productivity and social recognition give rise to two exchange rates: an Economic Exchange Rate (EER), capturing relative economic advantage, and a Social Exchange Rate (SER), capturing relative social visibility and recognition. AI strengthens the feedback between economic success and social standing, and the joint evolution of EER and SER is stable only when the product of two feedback parameters lies below unity. When this threshold is approached, the system enters a regime of systemic disequilibrium, in which economic and social disparities expand endogenously. Using panel data for 30 economies over the period 2012–2025, we provide empirical evidence of strong mutual reinforcement between economic and social advantage, with feedback strength rising as AI diffusion accelerates. The findings suggest that unequal AI access poses risks not only to equality but to social sustainability itself. The paper contributes a diagnostic framework for socially sustainable AI diffusion, highlighting the need for policies that dampen amplification mechanisms and strengthen inclusive pathways from economic performance to social recognition.