On Friday, Feb. 27, a student-led teach-in on the environmental impacts of AI and data centers was held in Dye Lecture Hall. The teach-in, which was led by College third-years Savannah Wright and Nick Barr Bono, discussed environmental consequences of AI technology and the various social factors that go into AI and data center development.
The event was organized by members of Students for Energy Justice, an environmentalist organization that has been involved in advocacy against pipeline developments and fracking. Wright noted that the group felt a need to put together an event about AI due to the increasing salience of AI-related issues and Oberlin College’s support for integrating AI into the academic environment.
“When we’re faced with a new technology that is kind of revolutionizing — or at least the narrative is that it is revolutionizing — education and communication and the way that we write and talk and obtain information, it’s important to think about where that comes from,” Wright said. “As Oberlin students, we have a responsibility to be aware of what we are interacting with. And that goes beyond just AI, that goes beyond a lot of other things. We are, as tuition-paying students, giving money to Oberlin, and we want to be aware of where that money is going. We also want to think critically about what kind of tools we’re using and how that might affect other people.”
A significant focus of the presentation was the discrepancy between claims of low water usage from AI companies and reports of massive amounts of water being used for AI. The presenters noted that while CEOs of AI companies often claim that their products use minimal resources, their methodology frequently does not take into account the costs of training their models, data center cooling, and resources expended by the power grid itself.
College second-year Afrina Maruf, one of the SEJ members who helped organize the teach-in, said that there is often very little transparency with regards to the energy and resources used by AI, making it difficult to calculate the exact amount of resources consumed by them. She said that while Sam Altman, the CEO of OpenAI, claims that each ChatGPT query costs only one-fifteenth of a teaspoon of water, the actual amount consumed is significantly higher due to the varying complexity of queries and the other factors of AI technology.
The presentation also included mention of the New Russia Township megasite, a recently announced development in a rural area north of Oberlin. While it is not clear exactly what is planned to be built on the site, the leaders of the teach-in noted that many local residents fear that a data center is planned.
“It’s especially … predatory because they were planning it for five years before … even releasing it to the public,” Maruf stated. “So then, it kind of gives the impression that no matter what the public’s opinion is on it, it’s going to be built, and with very little … information [from] the government. There have been megasites built before where the government says, ‘Oh, we’re not going to do too much development,’ … and they’ll have like three data centers and a million other things that people don’t want on their land.”
The presentation also noted the effects of data centers more broadly, explaining that many negative impacts on people and the environment often come with their construction. The town of Granbury, TX was used as a case study, with presenters describing how residents claimed that the amount of noise pollution and general health issues significantly increased after the construction of large data centers nearby. They also noted that the builders of data centers frequently try to convince local governments in less affluent areas to allow them to build by claiming massive economic benefits and a possibility of “missing out” on a great new technology.
“I feel like everybody needs to care about this,” College fourth-year Abby Fiedler, who attended the teach-in, said. “I care a lot about learning about AI because I hate it, and I want to know about it and its effects on the environment and jobs and people. … I think teach-ins are a good opportunity, especially for students, to have the opportunity to practice presenting this information and to educate. … I don’t always get exposed [to this information] if I’m not going out of my way to learn about it, so this is a perfect opportunity to learn.”
Alongside the free food from Lorenzo’s Pizzeria that was provided to all of the attendees, small flyers were given out that detailed ways for attendees to turn off or limit AI features on their devices and programs.
