Each day, about 2.5 billion prompts are submitted to ChatGPT, with it quadrupling its weekly annually. And with higher use of artificial intelligence and large language models such as ChatGPT, increasing amounts of energy and water are needed to sustain them.
Water is vital for LLMs for various reasons. LLMs use large data centers which require vast supplies of water for cooling. On top of that, they also need water for training AI and prompt responses.
Jenna Mu, a junior at The Kehillah School and member of Palo Alto’s Student Climate Coalition, said the use of generative AI models and the resources required for their operations disproportionately affects disadvantaged populations.
“The large data centers that help run AI are often being put in areas that were historically redlined and in usually lower income communities,” Mu said. “ All of the processing wastes energy, and this can deplete water from communities who really need it.”
Despite this, Mu said AI has the potential to make up for these environmental effects.
“AI is an incredible tool for researching possible sustainable energy and scientific innovation in general that would be really beneficial to combating climate change,” Mu said.
But Mu said AI and LLM’s are currently doing more harm than good for the environment.
“As it is currently being used right now and currently being trained and run, there is really just a disproportionate impact on just energy use, water usage and basically all the kinds of tangible things that it takes to run a large language learning model,” Mu said. “It’s really hitting hard currently, at this very moment, communities who are already disproportionately impacted by climate change.”
Despite the harms, senior and President of the AI Club, David Wu said the benefits AI and LLMs provide for innovation cannot be ignored.
“I would see it more as a tool for progress,” Wu said. “I think that there’s a lot of beneficial applications of AI, for example, in the medical space, in the legal space. People are developing ways of making it more energy efficient.”
Wu also said AI and LLMs can be helpful tools for learning.
“In Silicon Valley and in industry in general, there’s a lot of ways to use AI to increase productivity,” Wu said. “So, you know, humans don’t have to do a lot of basic tasks, and instead, can move on to, you know, higher level or creative tasks.”
However, Mu said she hopes high school students will be more considerate with their use of AI and LLMs because of the consequences that they can have.
“It’s really good to be conscious of AI use, especially for things such as schoolwork, where it’s something like, for example, you have a math assignment that you want to finish, but you don’t really want to look in the textbook to figure it out,” Mu said.
Senior Dylan Liao, who has used AI to help organize his work, agrees. He said learning about the environmental effects of AI and LLM has made him rethink how he should use it.
“It changes the way you look at it,” Liao said. “I didn’t realize it uses that much water, but if it uses that much water, then maybe I’ll look into using other sources, or maybe just trying to find sites like Reddit or like Quora, to answer my questions instead of using AI.”
Ultimately, Wu said he hopes governments restrict locations where data centers can be built.
“When you’re building the data center, you want to make sure that it doesn’t limit the local community’s access to water and electricity,” Wu said. “I think that this is something that maybe we could have some regulation (around) to make sure that we’re not harming the communities around them when we build data.”
