5110MAPA Creative Collaboration
AI and water: Can AI save water regardless of its energy usage?
AI is now something we use often, whether that be for academic or personal reasons - it is a lot of peoples first call aside from trusty google. A typical AI data centre can use upwards of about 3000,000 gallons of water a day - that is equal to the usage of about 1,000 different households whilst some data centres even using up to 5 million gallons per day (Brookings).
Across the globe the United Nations estimate that around 4 billion people are experiencing severe water scarcity for at least one month out of the year whilst 720 million others are living in countries with critical water supply. As climate change intensifies over the years these figures are expected to expand. Globally, agriculture accounts for around 70% of freshwater withdrawals and this puts immense pressure on our rivers, lakes and groundwater.
AI is now being used in places such as India and California to help measure monsoons and the groundwater depth, this in turn has helped thousands of farmers. This is being done through predicting droughts through machine learning models that can analyse rainfall patterns, detecting leaks and monitoring water quality. All of this is extremely helpful in regions that have unpredictable and difficult to understand climates however - AI uses a lot of water.
Now does this mean AI is both part of the solution and the problem? yes. This leaves us in a confusing situation morally and personally as it's confusing to think about how AI can both help a problem however, be a part of the problem itself! The conclusion is that AI is a broad spectrum with powerful tools that help us but to ensure it truly does benefit our water supply we must use it responsibly, transparently and with a deep awareness of our own environmental footprint.

