Year: 2025
Type: article
Abstract: The increasing integration of Artificial Intelligence (AI) into global economic activities raises significant questions about its environmental impact, particularly in resource-intensive sectors such as agriculture. This study utilized an Environmental Cost-Benefit Analysis (ECBA) framework to assess whether the economic benefits of leading AI models outweigh their ecological costs. We quantified energy consumption, carbon emissions, and water use based on real-world data from state-of-the-art models, monetizing these impacts using global externality values. Our analysis revealed significant differences in environmental efficiency: smaller models demonstrate clear net benefits, whereas large-scale models, such as Grok-3, incur extremely high ecological costs, justifiable only by exceptional economic returns. These findings suggest that the rapid expansion of AI is not compatible with sustainable development. Focusing on agriculture, where resource constraints are severe but AI holds the potential for increased productivity and resilience, we argue that implementing smaller, energy-efficient, domain-specific AI models can facilitate digital transformation without exacerbating environmental pressures. This research underscores the crucial need to integrate robust sustainability principles into AI design, deployment, and policy, particularly in ecologically sensitive sectors such as agriculture. This alignment is crucial to ensure that technological innovation and environmental responsibility work in tandem for a sustainable AI future.
Institution: King Saud University
Language: English
Topic: Green IT and Sustainability
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4414818019