The Environmental Cost of LLMs in AIED: Reporting and Practices

The article discusses the environmental and computational costs associated with the use of Large Language Models (LLMs) in the Artificial Intelligence in Education (AIED) sector, highlighting a lack of standardized reporting practices. It proposes an open-source framework for measuring and reporting these costs, including software solutions to assess the carbon footprint of both local and cloud-based hardware and a formula for calculating computational expenses of LLMs. This initiative aims to enhance transparency and ethical considerations in LLM deployment within the AIED community, which is crucial for practitioners to understand the hidden impacts of their models.