SustAIn: The Sustainability Index for Artificial Intelligence

How sustainable is Artificial Intelligence? At this point in time, we are aware that a lot of energy is needed in the development and application of AI, that so-called click workers classify data sets for training AI systems under very poor working conditions and that these systems quite often reinforce existing patterns of discrimination. In view of an increasing number of AI systems in use, there is an urgent need to discuss how AI can be made more sustainable. This is where our project SustAIn comes in.

From our perspective, we can speak of sustainable AI when the development and deployment of these systems respects planetary boundaries, does not reinforce problematic economic dynamics, and does not endanger social cohesion. We are thus considering the three central pillars of sustainability: Environment, Society and Economy. Sustainable economic development can only take place in a socially just society and both are tied to the protection of planetary boundaries. In addition, SustAIn is based on the 17 Sustainable Development Goals (SDGs) adopted by UN member states in 2015.

As part of SustAIn, we are working on tools that can be used to assess the sustainability of AI systems in practice. We show how AI can be made more sustainable and make related policy recommendations. The basis for this work is a set of criteria and indicators we have developed to assess the sustainability of AI. Our goal is to raise awareness among organizations, to provide them with tools to report about the sustainability of their AI systems and to make their AI systems more sustainable. In case studies, we will examine AI applications in sectors such as energy, mobility, and online consumption that are particularly relevant to sustainability and assess what potential they have to contribute to greater sustainability.

With SustAIn, we exploit the sustainability potential of AI in practice. Therefore, we seek exchange with organizations, developers, users, etc. in regularly held Sustainable AI Labs to inform about the sustainability of AI and to test our tools for their practical suitability. Once a year, we publish a Sustainable AI Report (for the first time in 2022).

Publications

Rohde, F.; Wagner, J.; Reinhard, P.; Petschow, U.; Mayer, A.; Voss, M.; Mollen, A. (2021):
​Nachhaltigkeitskriterien für künstliche Intelligenz. Entwicklung eines Kriterien- und Indikatorensets für die Nachhaltigkeitsbewertung von KI-Systemen entlang des Lebenszyklus

(Sustainability criteria for Artificial Intelligence. Developing a set of criteria and indicators for measuring the sustainability of AI along its life cycle) Schriftenreihe des IÖW 220/2021, Berlin (PDF)

Rohde, F., Gossen, M., Wagner, J., & Santarius, T. (2021):
Sustainability challenges of Artificial Intelligence and Policy Implications.

Ökologisches Wirtschaften (Ecological Economics), Journal, 36(O1), 36-40. (PDF)

Project Manager

Anne Mollen

mollen@algorithmwatch.org

Project partners