Asking ChatGPT 20 to 50 questions is equivalent to using half a liter of water. A study by the University of California, Riverside Research Center has estimated for the first time the water footprint of executed AI queries.
Data centers are now busy 24 hours a day processing the calculations needed to answer the many questions to AI services such as OpenAI, ChatGPT and Google’s Gemini. According to the published research by Riverside, writing just one command (prompt) per week by each of the 100 million users of the virtual assistant ChatGPT would lead to the consumption of five million liters of water per week. Most data centers use water cooling systems to prevent their servers from overheating.
FROM AWARENESS TO INNOVATION
The water footprint of AI models must be addressed as a collective top priority before it has an irreversible impact on the global water supply. There are plenty of examples of companies that have implemented innovative eco-solutions to make their systems more efficient and reduce their water impact. The big tech companies are also committed to this goal and have communicated their ambition to return a greater amount of water than they use in their direct operations by 2030.
WATER USE FROM AI: PROCESS OPTIMIZATION
Full awareness of your water footprint is the first step in implementing efficiency solutions. It is essential to use measuring instruments that meet high safety and reliability standards. Think of smart flow meters to monitor water consumption in real time, water leak detectors to quickly identify any inefficiencies and minimize waste, and probes to analyze water quality to prevent potentially dangerous contamination of equipment and operators.
