Toward Responsible AI Use: Considerations for Sustainability Impact Assessment

AI's resource consumption is a significant sustainability concern: The paper emphasizes that as AI and Large Language Models scale up with massive datasets, they consume "undeniably limited natural resources," highlighting that environmental impact is a critical consideration in responsible AI development that stakeholders across the AI industry need to address.

A new assessment tool can help measure AI's environmental and societal impact: The ESG Digital and Green Index (DGI) provides a dashboard for companies to assess and quantify their performance on sustainability metrics, including monitoring water and electricity consumption related to AI technologies, offering a practical mechanism for tracking environmental responsibility.

Sustainability assessment should align with global development goals: The DGI framework creates incentives for companies to align their AI practices with the United Nations' Sustainable Development Goals (SDGs), suggesting that responsible AI development should be connected to broader societal and environmental objectives rather than treated in isolation.

Multiple stakeholder perspectives are essential: The research brings together experts from AI, sustainability, human-computer interaction (HCI), and legal fields, indicating that assessing responsible AI requires interdisciplinary collaboration and input from diverse stakeholders across the AI value chain, not just technologists.

Current measurement approaches have limitations: The paper acknowledges "challenges and limitations" in their methodology, suggesting that while tools like the DGI represent progress, measuring AI's full environmental and societal impact remains an evolving challenge without perfect solutions.