The Environmental Cost of LLMs in AIED: Reporting and Practices

Researchers at AIED 2025 found that while most AI in education papers use Large Language Models, few report computational costs and almost none address environmental impacts. The study proposes open-source methods and software tools to standardize measurement and reporting of carbon footprints for LLM-based educational systems, addressing a significant transparency gap in the field.

The proliferation of LLMs in educational technology has created a hidden sustainability crisis that the research community has largely ignored. This study reveals a critical accountability gap: despite widespread LLM adoption in AIED systems, researchers systematically fail to measure or disclose the computational and environmental consequences of their implementations. This absence of reporting standards makes it impossible for institutions to assess the true ecological cost of deploying these technologies in educational settings.

The environmental cost of LLMs stems from their massive computational requirements during both training and inference phases. As educational institutions increasingly integrate LLMs into learning platforms, the aggregate energy consumption becomes substantial yet remains invisible in academic discourse. This mirrors broader industry trends where sustainability considerations lag behind adoption rates. The research community's silence on this issue enables a false narrative that technological advancement can proceed without environmental reckoning.

For educational institutions and EdTech developers, this work carries practical implications. Schools currently cannot compare the environmental footprint of different LLM-based solutions when making procurement decisions. Publishers and platform developers lack incentives to optimize for efficiency when environmental metrics go unreported. The proposed standardized measurement framework could shift market dynamics by making sustainability a competitive differentiator rather than a hidden externality.

The path forward involves adoption of these reporting standards across the AIED community. If researchers begin systematically disclosing carbon footprints alongside performance metrics, environmental considerations will gradually influence technology selection and development priorities. This normalization of sustainability reporting could establish precedents for other AI-intensive fields facing similar accountability gaps.

• →AIED conference papers predominantly use LLMs but fail to report computational or environmental costs, revealing a significant transparency gap.
• →Researchers propose open-source tools to standardize carbon footprint measurement for both local and cloud-based LLM deployments in education.
• →The lack of reporting standards prevents institutions from comparing environmental impacts when selecting AI educational systems.
• →Standardized environmental reporting could transform sustainability from a hidden externality into a competitive factor in EdTech markets.
• →The study highlights how accountability gaps in emerging technology fields enable unchecked environmental costs despite institutional values around sustainability.