Responsible Data Stewardship: Generative AI and the Digital Waste Problem

Generative AI creates a "digital waste" problem that goes beyond energy consumption: While researchers have focused on the energy costs of training and running AI models, the paper identifies a critical but overlooked issue—the environmental impact of storing massive amounts of synthetic data (text, images, audio, and video) that may never be used or serve no clear purpose.

Data storage itself is an ethical responsibility, not just a technical one: The paper argues that responsible AI development must address what happens to data after it's created, positioning indefinite data storage as an ethical issue comparable to concerns about bias and privacy that already receive significant attention in AI ethics discussions.

Other industries have solutions the AI community can learn from: Rather than starting from scratch, the paper recommends that AI researchers examine how other disciplines already manage digital waste and adapt those proven approaches—suggesting that solutions exist but haven't been systematically applied to generative AI yet.

Environmental justice across generations should be part of AI ethics: The paper expands the definition of responsible AI to include "intergenerational environmental justice," meaning that decisions about storing AI-generated data today have consequences for future generations and should be considered alongside current concerns like fairness and privacy.

Addressing digital waste requires research, technical tools, and cultural change: The paper calls for a multi-pronged approach including new research directions, specific technical interventions to reduce storage needs, and shifts in how the AI community thinks about data stewardship—suggesting no single solution exists.