Nudging users toward a greener GenAI – Scholar Q&A with Eugene Cho Snyder

With every advancement in the technology, artificial intelligence’s appetite for energy increases. A team of Page Center scholars is exploring the environmental impacts of generative AI, particularly its consumption of water and energy to operate. First-time scholars Eugene Cho Snyder, New Jersey Institute of Technology; and the University of Georgia’s Mengqi (Maggie) Liao and Nicholas Eng; and Ruoyu Sun, second-time scholar, are leading a three-part study that includes user and industry interviews and interface design.

The project is part of the Page Center’s 2025 research call on the ethics of generative AI. In this Q&A, Cho discusses the project’s potential to influence user behavior and industry practices and reveal practical solutions for reducing GenAI’s environmental footprint.

What is the origin story of your research idea?

I was inspired by a keynote by Kate Crawford at a tech conference called CHI in 2024, who mentioned the exploitation of natural resources by tech companies just to support their major AI developments. Those resources include electricity and minerals, which are obvious, but water too. I thought, humans are made of water, and now AI is drinking it all up? As more people use GenAI technologies like ChatGPT, it’s going to create serious resource issues. So, I thought maybe we should start looking at this from the individual user’s side, which could create meaningful and positive societal changes when collectively embraced by the masses.

How did the team get together?

When I got back from that tech conference, I pitched my idea at a summer research retreat facilitated by the Bellisario College of Communications and College of Information Science and Technology at Penn State. At the time, no one seemed interested, as more people were focused on studying GenAI use, let alone limiting its use. Maggie was a virtual attendee to that retreat, and reached out to me a few months later, saying, “Are you still doing this?” She mentioned the research call and her connections with public relations researchers at the University of Georgia, so we decided to get the group together and build something up. I’m very glad she did. By the way, three members of our team are proud alumni of Penn State and the Donald P. Bellisario College of Communications.

What’s your plan for the project? Where does everyone fit?

Maggie and I have the expertise in human computer interaction. Ruoyu and Nicholas are public relations scholars. I think this qualifies us as a pretty strong interdisciplinary team, although most of us graduated from the Bellisario College. There are three legs to this project. Ruoyu will be leading a public survey to understand public awareness of generative AI’s environmental cost. Once public awareness has been established, we can identify factors that predict awareness, and subsequently whether awareness of the environmental impacts of GenAI leads to an increase in environmentally-conscious AI behavioral intentions. After identifying public awareness, we then turn to industry opinions on this issue. Nicholas is spearheading this effort, conducting interviews with professionals in creative industries like advertising and public relations to better understand how they use GenAI in the workplace and what corporate environmental responsibility concernsthey may have. Lastly, Maggie and I will develop and test practical user-focused interventions, and see which kinds of interface design can promote pro-environmental choices when people are interacting with GenAI applications like ChatGPT.

Can you tell us more about the intervention?

The idea is to embed “green nudges” within the user interface, drawing from behavioral economics principles. Rather than forcing users to change or overwhelming them, the intervention aims to embed little cues that are not going to disrupt the design, but nudge people toward something more pro-social. For instance, the system might present internal messages within the dialogue to suggest more energy-efficient prompting behaviors, such as shortening prompts, or alternatively combining questions. We’re going to place these nudges within the dialogue that suggests more effective, energy efficient prompting behaviors.

Can you explain why studying the user is not only unique, but important?

When I looked at the literature, most research was at the beginning stage of trying to better estimate the carbon footprint of major GenAI services like ChatGPT. Or scholars would ask: “How do we make policy changes to control these data centers?” “How do we allocate computing resources more equally across regions and countries?” They are based on very technical and macro-level approaches. We will fill in the gap of user-centered approaches. I think we're trying to be academic advocates, helping promote more environmentally conscious use of GenAI among users. That's the differentiation.

Share the Page Center’s role in the project? How does the grant help make this project happen?

This is an interdisciplinary collaboration between two schools and two disciplines. Without this type of grant, it can be tricky and difficult to allocate human resources and finances to maintain a multi-site project. Even when I pitched the idea, I was thinking of a pilot study. With this grant, it grew into a more comprehensive and systematic project. I didn’t expect this, but as I invest more in the project, the more I’m convinced that this shouldn’t be a one and done. If our intervention is successful, for instance, we can revise the prototypes and push them to be tested for their longitudinal effects. Thanks to the support, there’s momentum and I want to share my appreciation for the Arthur W. Page Center.