AI being thirsty? Study says ChatGPT drinks 500ml of water per conversation

OpenAI's ChatGPT is all over the internet since its launch in 2022 with discussions around the grand potential of artificial intelligence and the threats it poses. Notably, in all the conversations around the potential threats of the ChatGPT-like AI platforms, very few are noticing the environmental impact. A new study published by the University of Colorado Riverside and the University of Texas Arlington reveals the water footprint of AI models and it was revealed that “ChatGPT drinks 500ml of water” for every conversation. 

The ChatGPT has gained immense popularity in the past few months with billions of people using the platform daily. The water footprint of the AI platform is alarming as it tells about the daily consumption of fresh water.

The study released that training GPT 3 at Texas Microsoft’s state-of-the-art US data center consumes nearly 7,00,000 liters of water and the numbers are expected to be three times higher in the data centers of Asia, the paper named “Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models,” said which is currently awaiting peer review.

The study keeps its focus on ChatGPT, but in reality, OpenAI has triggered a race among technology companies toward the most sophisticated AI platform. Almost all big tech firms are currently working on their AI modes like Google Bard, Meta's Llama, etc. The logic goes the same for these AI platforms also as the process to train these AI models is almost the same.

How AI tools consume water?

The water footprint of an AI model like ChatGPT is primarily associated with the energy consumption required to power the servers and data centers that host and run the model.

Data centers, where AI models are stored and processed, require cooling systems and energy-intensive computing resources, which can contribute to water usage through cooling water and electricity generation. The water footprint of AI models, therefore, depends on the energy mix and cooling methods used by the data centers, as well as the efficiency of the hardware.