The hidden cost of AI's for natural enviroment
While generative artificial intelligence predates ChatGPT, this tool, developed by OpenAI's engineers, definitely skyrocketed the popularity of this technology. It quickly garnered substantial attention and has profoundly impacted how we study, acquire information, or generate various types of text.
ChatGPT eliminates the need to spend time seeking information or creating content. But this convenience doesn't come without a cost. The processes carried out by generative artificial intelligence lead to massive energy consumption and substantial heat production. As is the case with humans, the heat increases the tool's "thirst". It appears that ChatGPT has an unusually large thirst.
ChatGPT's water consumption
Researchers from the University of California approximated that ChatGPT needs about a cup of water to answer at most 25 questions. That implies around 0.26 gallons of water for every 100 questions, primarily used to cool data centers. While this amount might not seem enormous, it rapidly grows when we consider the actual number of processed questions. Depending on the source of calculation, ChatGPT processes between 10 million to 1 billion questions daily.
Furthermore, the rising water usage in the tech industry hasn’t gone unnoticed by environmental experts. For instance, Microsoft's water consumption for cooling its data centers surged by 34% between 2021 and 2022. At present, it hovers around a staggering 1.7 billion gallons.