ARTIFICIAL INTELLIGENCE FOR ENVIRONMENTAL SUSTAINABILITY: CONCEPTUAL FOUNDATIONS, THEORETICAL PERSPECTIVES, AND A FRAMEWORK FOR BALANCED ASSESSMENT

Abstract

Artificial intelligence (AI) is increasingly positioned as a critical instrument for addressing environmental degradation, climate change, and resource depletion, while simultaneously being implicated as a growing source of energy consumption, carbon emissions, and material extraction. This conceptual review critically examines AI's dual role as both an environmental solution and an environmental burden, integrating theoretical perspectives from environmental studies, science and technology studies, sociology, economics, and governance scholarship. Rather than cataloguing technical applications, the review foregrounds how foundational theories, including ecological modernization theory, socio-technical systems theory, the planetary boundaries framework, circular economy theory, the triple bottom line, diffusion of innovation theory, responsible innovation theory, and risk society theory, help explain why AI's environmental value is conditional rather than automatic. The review synthesizes recent empirical and conceptual literature on AI's applications across climate, energy, agriculture, biodiversity, water, waste, and urban systems, alongside literature documenting AI's energy, carbon, water, and material footprints. It concludes that AI functions as a capable but conditional sustainability tool whose net environmental value depends on governance quality, measurement rigor, and the extent to which efficiency gains are prevented from generating rebound effects. A conceptual framework linking AI technologies, environmental applications, governance mechanisms, and sustainable development outcomes is proposed to guide future interdisciplinary research.