Introduction: As institutions increasingly hand high-stakes decisions to algorithms—who receives a scholarship, a loan, or a job—they also cede a share of the public trust on which cohesive communities depend. This study asks whether artificial intelligence strengthens or weakens the institutional standing and collective belonging that allow communities to function, using scholarship allocation as a case study where algorithmic decisions shape who is included, recognized, and treated as deserving.

Background: Drawing on Ibn Khaldun's concept of asabiyyah (group solidarity), we define social cohesion as the bond between human dependence and the institutions that organize shared needs into collective life—sustained by leadership that fulfills its obligations and preserves the conditions under which solidarity endures, especially in times of crisis. Algorithmic bias matters here because it can disrupt those conditions, altering how citizens interpret the fairness of public decisions.

Methods: To make this mechanism explicit, we present a Monte Carlo simulation modelling students who compete for a limited number of scholarships (the top 20% of applicants) allocated under two conditions: a fair system scoring applicants on ability alone, and a biased system that systematically penalizes one group. We translate disparities in scholarship allocation into a composite social cohesion index based on trust, participation, and cooperation.

Results: Across 1,000 repetitions, the fair system produced a small fairness gap (0.021) and higher cohesion (73.1), while the biased system produced a larger gap (0.191) and lower cohesion (57.9).

Discussion: Because the fairness–cohesion relationship is assumed rather than estimated, the model's value lies in exposing its assumptions to critique and generating a testable hypothesis: widening algorithmic gaps predict declining cohesion.

Conclusion: We propose to evaluate this empirically through surveys, administrative records, platform-level data, and interviews, contributing a conceptual and empirical roadmap for understanding asabiyyah in the age of AI.

Limitation: A limitation is that the relationship between algorithmic fairness and social cohesion is specified within the simulation rather than empirically estimated.