Governing AI for ESG and SDG Outcomes in the Global South: a Systematic Review

The systematic review examined 30 peer-reviewed articles published between 2020 and 2026 concerning the application of Artificial Intelligence (AI) in supporting Environmental, Social, and Governance (ESG) practices and Sustainable Development Goals (SDGs), particularly within the Global South. The research sought to identify which AI applications support ESG performance and SDG outcomes, how technical, institutional, and sociocultural factors influence their adoption, and the governance mechanisms for their responsible integration. Findings indicate that evidence is concentrated in environment-related SDGs, where predictive analytics and AI-enabled remote sensing aid in climate risk assessment, urban planning, and land-use monitoring. Successful integration of AI into sustainability initiatives requires bundled interventions involving data infrastructure, institutional capacity, participatory design, and accountability safeguards. Furthermore, social and governance applications are heavily contingent upon mitigating algorithmic bias, data inequity, and exclusion. Governance responsibility is distributed among governments, corporations, civil society organizations, local communities, and international development partners.

The pattern emerging from the literature is that technological efficacy in achieving sustainability goals is fundamentally mediated by context and governance structures, especially in the Global South. The concentration of evidence in environmental SDGs suggests a current focus on tangible, data-intensive applications like remote sensing, while social and governance aspects introduce complex relational challenges concerning equity and bias. This suggests a latent tension between the technical potential of AI (e.g., predictive analytics) and the institutional realities of developing contexts. The proposed Context-Responsiveness Matrix points toward a necessary shift from viewing AI as an inherent technological effect to understanding it as a sociotechnical accomplishment, demanding context-specific governance tailored to local institutional capacity and problem complexity. The distribution of governance responsibility across multiple actors implies that technical deployment alone is insufficient; the core challenge lies in establishing equitable accountability frameworks that recognize and address existing social and institutional asymmetries among various stakeholders involved in the AI ecosystem. What infrastructure needs to be established to ensure these distributed responsibilities translate into equitable outcomes rather than reinforcing existing inequalities? What mechanisms must be prioritized to move beyond mere adoption toward genuinely responsible integration across diverse global settings?