Artificial Intelligence and Social Inequality: A Review of Algorithmic Bias, Digital Exclusion, and Governance Responses

Artificial intelligence (AI) is increasingly integrated into employment, education, healthcare, financial services, public administration, and social welfare systems, creating opportunities to improve efficiency, accessibility, and decision-making. However, the growing reliance on AI also raises concerns about social inequality, particularly when algorithmic systems reproduce existing patterns of discrimination, exclude marginalized populations, or concentrate technological benefits among already advantaged groups. This review examines the relationship between AI and social inequality, focusing on algorithmic bias, digital exclusion, and emerging governance responses. It synthesizes conceptual and empirical literature on discriminatory data practices, unequal access to digital infrastructure, disparities in AI literacy, and the socioeconomic consequences of automated decision-making. The review also examines how inequalities associated with gender, race, socioeconomic status, disability, geography, and language can influence the design, deployment, and outcomes of AI systems. The literature indicates that algorithmic discrimination may arise from unrepresentative datasets, historical inequalities embedded in training data, inappropriate performance measures, and insufficient institutional accountability. Digital exclusion further limits the capacity of disadvantaged populations to benefit from AI-enabled services, particularly in contexts characterized by inadequate connectivity, limited digital skills, affordability constraints, and linguistic barriers. Although regulatory frameworks, algorithmic audits, inclusive design, and human oversight offer important safeguards, their effectiveness depends on implementation, transparency, and meaningful participation by affected communities. The review argues that addressing AI-related social inequality requires an integrated approach combining technical fairness, digital inclusion, institutional accountability, and rights-based governance. It concludes that AI should be evaluated not only in terms of technological performance but also according to its contribution to social justice, equitable access, and inclusive development.