The use of artificial intelligence (AI) is transforming our everyday lives, from virtual assistants to advanced machine learning tools. However, This technological revolution has a significant environmental costparticularly in terms of water.
ChatGPT’s water consumption is linked to the data centers where they operate. These technological giants, which process enormous volumes of information, generate a considerable amount of heat, which requires cooling systems intensive to keep the servers in optimal conditions.
However, the problem arises when Most of these systems must use potable water, as impurities from untreated water or salt and other components can damage servers’ sensitive electronic systems. In addition, many centers use methods of evaporative coolingwhere water is not reused, further increasing the impact.
How much water does Chat GPT consume?
According to a group of researchers from Riverside University (California) and Arlington University (Texas), Asking between 10 and 50 questions to ChatGPT-3 can require up to two liters of water. That is, what a human being should drink per day to stay healthy. Not only that, Research has also proven that the amount varies in relation to the complexity of the queries and with the arrival of more advanced models, such as GPT-4, the calculations necessary to provide more precise answers also increased.
A report from The Washington Post details that GPT-4 consumes approximately 519 ml of water to generate a 100 word emailwhich is equivalent to a small bottle. Likewise, the Spanish Engineering Institute revealed that a ChatGPT query requires three times more energy than a Google search. This level of consumption, multiplied by millions of users, generates an alarming cumulative impact on the global water supply.
Measures to mitigate the water footprint
Despite the outlook, some technology companies are taking measures to reduce the water footprint of their operations. Google, for example, has implemented artificial intelligence systems to optimize the cooling of its data centers, achieving a 40% reduction in energy consumption. Other companies, such as Digital Realty, they have started using non-potable waterlike rainwater, to cool your facilities.
In addition, innovative technologies are being developed, such as Iceotope’s “precision liquid cooling”, which completely eliminates the need for water in server cooling. These initiatives represent big steps towards greater sustainability in the technology sectoralthough there is still a long way to go.