Embedding Environmental Sustainability in GenAI Usage
A design science approach to explore interventions for sustainable GenAI interaction
A.M.W. van Laarhoven (TU Delft - Technology, Policy and Management)
J Ubacht – Mentor (TU Delft - Information and Communication Technology)
ME Warnier – Mentor (TU Delft - Multi Actor Systems)
Erik Vermeulen – Mentor
Jasper Snijder – Mentor
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This thesis investigates how organisations can embed environmental sustainability within the operational use of Generative Artificial Intelligence (GenAI) tools through targeted, user-centred interventions. While research has largely focused on reducing emissions in AI model training, the inference phase, where GenAI is integrated into daily workflows, remains a substantial and under-addressed source of environmental impact.
Using a Design Science Research approach, the study combines a literature review, interviews with GenAI users and AI experts, and behavioural theories including the COM-B model, Theory of Planned Behaviour, Nudging, and Affordance Theory. Enabling factors for pro-environmental GenAI use were translated into functional and non-functional requirements, guiding the development of three persona-specific interventions: (1) Sustainable by Default for externally motivated users, embedding energy-efficient model settings and a monitoring dashboard; (2) Sustainability Guidance for aware but uncertain users, offering a sustainable prompt builder and impact estimator widget; and (3) Collective Sustainability for unaware users, providing monthly emissions feedback and rotating green tips.
The resulting integration framework and decision-support tool offer practical guidance for embedding sustainability into organisational AI practices, demonstrating that environmental impact reduction in GenAI requires socio-technical, behavioural, and cultural change alongside technical optimisation.