ARTIFICIAL INTELLIGENCE AND SOCIOECONOMIC INEQUALITIES: ALGORITHMIC BIAS, LABOR MARKET IMPACTS, AND GOVERNANCE AGENDAS FOR EMERGING ECONOMIES

Abstract

The popularization of generative artificial intelligence (AI) after 2022 has intensified debates about algorithmic bias, transparency, privacy, accountability, and social justice. This study conducts an integrative review with a focus on evidence-based governance, comparing recent regulatory frameworks (European Union AI Law, General Data Protection Law, and PL 2338/2023 in Brazil), risk management frameworks (NIST AI Risk Management Framework), and multilateral recommendations (UNESCO and OECD). The results of reports on the effects on work, productivity and inequalities were systematized, including documented cases of algorithmic bias such as Amazon (2018) and COMPAS (2016). Brazil emerges as a regional leader in AI adoption, with 40% of companies using the technology systematically, but investing only 30% of their economic potential. A public policy framework for emerging countries is proposed structured in five pillars: risk-based regulation, algorithmic impact assessment, institutional strengthening, labor and retraining policies, and transparency with social participation. The results indicate that risk-oriented regulations, combined with management structures and algorithmic education, tend to reduce harms and amplify the social benefits of AI.

DOI: https://doi.org/10.56238/edimpacto2025.084-008