AI Sustainability Workshop

by Malcolm Campbell-Verduyn, Marco Zullich, Juan Diego Cardenas Cartagena, and Stefanie Kunkel

AI sustainability

The increasing scale, popularity, and widespread adoption of AI systems in the last few years stimulates urgent questions about sustainability. The energy footprint of the (hyper-scale) data centers powering AI systems is consuming a growing share of the overall energy consumption of countries and communities around the world, including in Europe and The Netherlands.

Environmentally sustainable AI

Environmentally sustainable AI refers to AI that is optimized to address (1) direct environmental effects, related to the AI systems’ life cycle energy and material requirements, such as greenhouse gas emissions in training, or hardware requirements, as well as (2) indirect environmental effects that arise when AI systems are applied in other application domains, such as industrial companies or private households.

Why sustainability matters in AI

The workshop begins by examining why sustainability matters in AI, followed by an introduction to its technical and socio-economic dimensions. Participants will then work in groups on a hands-on "Rebound Archetype" exercise, analyzing stakeholder scenarios to identify and address direct and indirect socio-environmental effects of AI.

Goals of the workshop

By the end, you will understand key sustainability challenges in AI, explore multiple perspectives, collaborate across disciplines, and learn a method for assessing the broader impacts of AI-related decisions.

This workshop is open to both university staff and students. It is part of a larger, wider discussion group on how to bring environmental sustainability into AI education.

_{This is a Research Support Hub workshop. The Research Support Hub is a CIT initiative to support all researchers and PhD-candidates throughout the University of Groningen on data management, geo-services, statistics, machine Learning, AI-tools and high performance computing.}