About the project
Aim
The aim of Green Teaming AI project is to develop theoretical concepts, and analytical devices, to 'test' and critique the ways in which ‘the environment’ is represented, rendered, and prefigured (or fails to be so) in GenAI. It develops Green Teaming AI as a participatory approach to address environmental concerns. It engages diverse stakeholders and raises public and research awareness of GenAI’s indirect environmental effects.
Method
Our research is broadly informed by new materialism and digital STS. Throughout the course of the project, a set of co-creative workshops will be arranged with diverse participants across Sweden. Each workshop focuses on specific contexts of human-environment interactions or sites of environmental meaning-making. Examples include, but are not limited to, economy, ecology, marketing, nature conservation, forestry, and tourism.
Green Teaming AI draws inspiration from the method of ‘red teaming’ AI. Red teaming is typically conducted in technology companies to identify unintended, unsafe, and harmful outcomes of AI models.
Significance
The Green Teaming AI approach stimulates thought, discussion, and action at the intersection of environmental concerns and AI systems by eliciting and problematising the embedded biases towards e.g. nature, species, or geographical locations. Through its stakeholder-engaged, workshop-based approach, this project creates and strengthens connections between society and research.
Researchers
Jutta Haider, Swedish School of Library and Information Science, University of Borås
James White, Department of Technology and Society, Lund University
Björn Ekström, Swedish School of Library and Information Science, University of Borås
Malte Rödl, Division of Environmental Communication, Swedish University of Agricultural Sciences
Contact
Green Teaming AI is part of Mistra Environmental Communication.