“`html

Generative artificial intelligence technologies consume significant energy, however, numerous AI creators are optimistic that this technology could ultimately serve as a benefit for the climate — potentially resulting in a more sustainable power grid, for example.

Nonetheless, the aims of individuals creating AI systems for environmental issues and those actively involved in climate action may not necessarily align. To assess the two factions, University of Washington researchers engaged with nine AI developers for sustainability — from a graduate student to a startup entrepreneur — and ten climate advocates, including a community activist and an employee of an environmental nonprofit. They discovered that while both developers and advocates were passionate about climate initiatives, their distinct values and views differed significantly, particularly regarding ethical considerations.

The team shared its findings on July 8 at the Designing Interactive Systems Conference in Funchal, Portugal.

UW News interviewed lead author Amelia Lee Doğan, a doctoral candidate at the UW Information School, along with senior author Lindah Kotut, a UW assistant professor in the Information School, about the research findings.

There’s significant concern regarding the environmental impact of training AI and deploying the models, but also a possibility for AI to make a positive contribution. How did you observe the perspectives of the different groups you interviewed diverging on these viewpoints?

Amelia Lee Doğan: Most advocates perceived AI as potentially beneficial in highly specific contexts. This encompasses automating existing tasks and bridging connections between community members and the natural environment, such as developing identities for natural entities like rivers or trees or creating diagnostic tools for urban agriculture. There were also a few traditional applications in climate science. However, some individuals expressed worries not just about the environment, but also about labor issues. AI won’t resolve challenges stemming from policy. In these scenarios, climate change isn’t the sole issue; it’s tied to broader injustices.

What insights did you gather from this investigation that took you by surprise?

ALD: We received numerous intriguing concepts from advocates and documented some in a fictional style for another publication released in January. This is what we term a design fiction. You envision a technological future and scrutinize it from an academic standpoint. Some ideas were remarkably innovative — such as, what if a river could articulate thoughts in English while also expressing itself through sounds like water bubbling or music?

Lindah Kotut: When we presented our findings at a conference, many attendees were taken aback that a sizable portion of the developers were unaware of the initiatives undertaken by grassroots activists. Conversely, the activists generally have some understanding of what the developers are pursuing. If the roles were reversed, I believe many discussions surrounding the implications of AI on the environment would be significantly more developed.

ALD: Numerous developers in research or nonprofit realms had noticeably less interaction with advocates compared to their counterparts in business. I posed a straightforward question to the developers: Have you communicated with anyone from the community you’re aiming to assist? Many answered in the negative. The advocates expressed a desire for these developers to participate in protests or meetings.

Did individuals provide suggestions for addressing this communication gap?

ALD: We suggest initiating discussions as a primary step. Advocates would appreciate being approached by developers to understand their needs and desires, with the recognition that advocates are often heavily resource-limited. For instance, one advocate we spoke with had been utilizing a non-AI tool created by the government. When maintenance for that tool was discontinued, it hindered her workflow. Thus, she showed interest in the development and upkeep of new software tools.

LK: Both sides face limitations. Certain developers are tackling environmental justice topics as side projects, which restricts the time they can devote to it. The advocates, in turn, are primarily accountable to communities and the environment and are often restricted by policy. Therefore, they concentrate on fostering changes in policy that will influence technology. In contrast, developers may find themselves in situations where corporate interests conflict with environmental objectives. We don’t have a clear resolution for that tension.

What aspects of AI technology were advocates hopeful about?

ALD: Much of the work done by advocates involves substantial data and could gain from automation — for example, sifting through government databases of PDFs that lack high-resolution scanning. Advocates are also enthusiastic about advancements in scientific research.

What do you wish for the public to understand regarding this research?

ALD: Climate change is an immediate issue, and we can’t necessarily wait for uncertain promises from AI. We already recognize that the most effective remedies for the climate emergency involve policy solutions: reducing fossil fuel usage and safeguarding our lands and waters.

We discovered that societal issues afflicting technology development similarly manifest in climate tech development. Many power dynamics come into play. Developers don’t always have the liberty to define the overarching vision for a project. This concern also applies to developers engaged in social good projects who can still fall into many of the traps that individuals in the tech sector encounter.

LK: Anyone able to do so should amplify the efforts of grassroots climate communities. This helps to disseminate the changes they advocate for in the manner they wish. Moreover, in technology, our pace is incredibly rapid. Move fast and break things has been a guiding principle. However, one of the best counterbalances to this fast-paced approach is to heed the voices of those you aim to support. Engage with local climate organizations and truly listen to them. Supporting these groups is a crucial step, but at times, simply listening represents the best course of action. These organizations have a deep understanding of the concerns faced by communities directly impacted.

 

Hongjin Lin, a PhD candidate
“““html
at Harvard University, is a co-author on the research. This study was partially supported by the National Science Foundation and the University of Washington’s Graduate School’s Office of Graduate Student Equity & Excellence.

For further details, reach out to Doğan at [email protected] and Kotut at [email protected].

“`