The use of AI technology has become increasingly prevalent in our daily lives, from chatbots to virtual assistants. However, what many people may not realize is the significant impact that AI has on our global water resources. According to recent studies, AI systems and large language models (LLMs) consume a significant amount of water, with some estimates showing that a single chat with an AI tool can use up to half a litre of water. This may seem insignificant, but when millions of people use AI on a daily basis, it adds up and contributes to the overall water footprint.
Research by experts in Information and Communications Technology (ICT) has shown that AI chatbots can consume up to 0.5 litres of water during each conversation with a user. This applies to all AI systems and LLMs in place. In addition, a study by the University of California, Riverside has projected that the demand for AI will lead to a significant increase in water withdrawal from freshwater sources, which could be equivalent to the annual water withdrawal of entire countries.
One of the main contributors to AI’s water usage is the data centres that house these systems. These centres require significant amounts of water for cooling purposes, as the high-performance servers generate a lot of heat. In addition, water is also needed for electricity generation in thermoelectric power plants, which provide energy for these data centres. The manufacturing of AI chips also requires large amounts of water, particularly in processes like water fabrication.
Overall, the constant thirst for water by AI systems is a growing concern, especially in the face of global water scarcity. As the use of AI continues to increase, so will its impact on our water resources. It is crucial for companies and researchers to consider the environmental impact of AI and find ways to reduce its water consumption.
