Abstract
The integration of Green Artificial Intelligence (AI) technologies into educational systems offers a promising avenue to enhance operational efficiency while addressing sustainability challenges. Through a rigorous three-phase methodology combining literature review, AI agent development, and participatory workshop-based case analysis, this paper highlights the pivotal role of AI agents, as applications of Green AI technologies, in driving transformative outcomes within schools. By directly improving self-learning efficiency and reducing learning costs for students, enhancing management and service efficiency, reducing labor costs for schools, as well as minimizing resource dependence for both teachers and students, AI agents create a foundation for sustainable operations. These direct effects generate positive spillover effects, cascading into broader outcomes, including innovation performance, economic efficiency, and environmental sustainability, aligning with the United Nations Sustainable Development Goals (SDGs). By presenting a comprehensive conceptual model, this study demonstrates the pathways through which Green AI contributes to sustainable development in education and emphasizes its critical role in bridging technological innovation with sustainability.
Introduction
While Artificial Intelligence is revolutionizing education, the "Red AI" trend—characterized by massive computational costs and high carbon footprints—threatens to undermine global sustainability efforts. Furthermore, sustainability in schools is often viewed through the narrow lens of policy or funding rather than technological integration.
This paper proposes a conceptual framework for transforming schools into sustainable institutions by deploying "Green AI" agents. By focusing on energy-efficient AI design and resource optimization, the authors demonstrate how these digital agents can simultaneously improve learning outcomes, reduce operational costs, and minimize environmental impact, aligning educational technology with United Nations Sustainable Development Goals (SDGs).
Methodology and Data
The researchers employed a three-phase methodology to build and validate their conceptual model:
- Literature Review: A systematic synthesis of existing research on Green AI, AI agents, and sustainable development in education.
- AI Agent Development: To ground their theory in practice, the team developed and deployed over 20 specialized AI agents. These agents were designed using Green AI principles (optimizing for energy efficiency and minimal computational waste) to handle tasks like academic advising, resource management, and personalized tutoring.
- Participatory Workshops: A case analysis phase involving workshops with over 2,300 users (students and teachers). These sessions gathered qualitative feedback on how the agents affected learning efficiency, labor costs, and resource dependence.
Instead of standard open-source datasets, the study utilized real-world user interaction data and qualitative findings derived from the participatory workshops where students and teachers evaluated the agents' impact on their daily workflows.
Key Insights and Conceptual Framework
The paper’s core contribution is a conceptual model that maps out seven hypotheses (H1–H7) linking AI applications to sustainability metrics.
Direct and Indirect Pathways
The model distinguishes between Direct Effects (H1-H5) and Indirect Pathways (H6-H7):
- The "Digital Extension" of Teachers: AI agents act as 24/7 virtual mentors. This student-centered approach compensates for resource limitations in traditional systems, particularly in under-resourced regions, thereby democratizing education.
- Efficiency and Resource Independence: By reducing paper waste, lowering labor costs, and improving self-learning efficiency, schools can drastically reduce their reliance on physical infrastructure and consumables, directly boosting environmental performance.
- Positive Spillover Effects: Efficiency gains in one area (like administrative automation) create "cascading benefits" in others, such as freeing up human teachers to focus on high-value pedagogical innovation rather than mundane tasks.
Results and Practical Outcomes
While primarily a conceptual paper, the pilot deployment yielded several practical outcomes:
- Technical Feasibility: Successfully served over 2,300 users with 20+ AI agents, proving the model's scalability.
- Operational Efficiency: Qualitative feedback indicated a significant reduction in mundane administrative tasks for staff and "learning costs" (such as tutors and physical textbooks) for students.
- Equity and Accessibility: The agents bridged the "digital divide" by providing instant access to quality academic guidance with minimal requirements (only electricity and internet), supporting educational equity.
Conclusion and Future Directions
The authors acknowledge that this work is foundational. Future research should focus on quantitative validation through large-scale, longitudinal studies to measure exact ROI and spillover effects. Additionally, addressing implementation barriers—such as cultural and institutional resistance—and developing standardized "sustainability impact metrics" for educational AI will be critical for long-term success. This framework provides significant theoretical insights for further empirical research while offering actionable strategies for policymakers and educators to harness Green AI for building sustainable schools.
Summary updated