Towards an Environmental Ethics of Artificial Intelligence

AI's environmental footprint has been largely ignored in ethics discussions: While AI ethics research has focused heavily on transparency, privacy, bias, and safety, the paper identifies a significant gap—environmental impact has "largely escaped ethical scrutiny" despite evidence that training and deploying AI systems requires substantial energy and resources.

Environmental justice requires considering three dimensions of fairness: The paper applies environmental justice frameworks to AI, arguing that ethical AI development must address distributive justice (who bears the environmental costs and benefits), procedural justice (who gets a say in AI decisions), and recognition justice (whose values and interests are acknowledged), expanding AI ethics beyond just human concerns.

AI decisions can harm ecosystems and non-human actors: The paper emphasizes that AI's environmental ethics must extend "beyond an anthropocentric focus towards including more-than-human actors such as animals and ecosystems," recognizing that AI outcomes can negatively impact the natural world, not just people.

We need new criteria for deciding when AI development is justified: The paper raises a fundamental question that currently lacks clear answers: "When is it (un)justifiable to develop an AI system?" This suggests we need explicit ethical frameworks to weigh whether an AI system's benefits justify its environmental costs before building it.

Design choices matter for environmental responsibility: The paper calls for considering environmental impact in AI system design decisions from the start, implying that how we build AI—not just whether we build it—should be guided by environmental justice principles.