Built a tracker to estimate water wastage when talking to Claude (github.com)

A new project called Sip has been launched to estimate water wastage associated with querying the Claude language model. This initiative addresses a critical but often overlooked aspect of AI operations—environmental impact. As users interact with the AI, Sip provides real-time updates on the estimated water consumption linked to each prompt, effectively turning abstract energy costs into tangible figures. For instance, the tool might display that a single query consumes approximately 1.2 mL of water, accumulating to a daily total that users can track seamlessly alongside usage statistics like session cost and context window percentage. This development is significant for the AI/ML community as it raises awareness about the resource usage inherent in machine learning, particularly as concerns about sustainability grow. By utilizing locally-based calculations that avoid extra API calls, Sip empowers users to monitor and manage their environmental footprint without impacting their operational costs. The technical underpinnings involve estimating water usage based on output tokens and energy consumption metrics, filling a crucial gap in understanding the ecological implications of AI technology in real-time and prompting a discussion on optimizing AI systems for greater efficiency.