According to University of California researcher Shaoli Wren, ChatGPT and similar LLMs use up to 500 milliliters (17 ounces) of water for every 20 to 50 prompts or questions users ask. in the newspaper Published on Arxiv Earlier this year, the researcher noted that 500ml might not seem like a lot, but with the combined use of ChatGPT by users around the world, the water usage is huge.

Talking to AP NewsRan was quoted as saying:

‘It’s fair to say that most [water] growth [in Microsoft’s 2022 environmental report] stems from AI. Its heavy investment in generative AI and partnership with OpenAI. Most people are not aware of using the resources underlying ChatGPT. If you are not aware of the use of the resources, then there is no way we can help conserve the resources.’

Microsoft acknowledged this problem when asked by AP News and said it is exploring ways to measure its energy use and carbon footprint. He also said it is looking for ways to make the LLMs less energy-intensive.

The company told AP News that:

“We will continue to track our emissions, accelerating progress while increasing our use of clean energy to power data centers, purchasing renewable energy and other efforts to meet our sustainability goals of being carbon negative, water positive and zero waste by 2030.”

OpenAI also acknowledged the water usage problem and said it was looking for ways to make the LLMs more energy efficient.

While the amount of water used to run an LLM is quite staggering, it is only fair to note that these modern LLMs are not that old, so over time methods can be developed to reduce water and energy consumption.

For most people, most people’s first interaction with generative AI was ChatGPT which came out in late 2022. According to GizmodoChatGPT has seen a drop in site visits for three consecutive months.

As the numbers accessing ChatGPT decrease, apparently, the technology’s water consumption also decreases.