Earth for AI: A Political Ecology of Data-Driven Climate Initiatives. Issue 130 (March 2022)

• Record Type: Journal Article
• Authors: Nost, Eric; Colven, Emma
• Is Part Of: Geoforum. Issue 130(2022)
• Page Start: 23
• Page End: 34
• Publication Date: 2022-03
• Subjects: Adaptation -- Artificial intelligence -- Climate change -- Digital geographies -- Environmental data justice -- Knowledge production; Geography -- Periodicals; Human geography -- Periodicals; Regional planning -- Periodicals; Sciences de la terre -- Périodiques; Géographie -- Périodiques; Géographie humaine -- Périodiques; Aménagement du territoire -- Périodiques; Earth sciences; Geography; Human geography; Regional planning; Periodicals; Electronic journals; 304.205
• DOI: 10.1016/j.geoforum.2022.01.016
• Languages: English
• ISSNs: 0016-7185

Abstract

Highlights: Environmental and climate crises are grist for tech solutions. AI technologies require computational resources that are environmentally embedded. Disasters become "shocks" that the AI industry capitalizes on. Climate AI shapes material investment flows and landscapes. Data-driven approaches to climate reproduce injustices faced by marginalized communities. Abstract: Emerging narratives around artificial intelligence (AI) and machine learning place great faith in these technologies' ability to ameliorate threats posed by climate change. They promise the capacity to analyze vast amounts of more precise and real-time data, improving how decision-makers predict, respond, and adapt. Yet scholars in political ecology have long observed that technocentric approaches typically reduce complex human-environment relationships in ways that fail to account for social relations and power dynamics. This paper charts the emerging political economy of "climate AI" – the philanthropies, NGOs, private consultancies, and tech giants investing in data-driven climate initiatives. Mapping out two case studies, we show that environmental and climate crises are grist for tech solutions and find that many climate AI actors are interested in it for surveillance, greenwashing, and commodifying algorithms. We pay special attention to how neocolonial and racialized power structures manifest in climate AI and outline three ways for political ecologists and digital geographers to research its socio-materiality: how computational resources are environmentally embedded, how disasters become "shocks" that the AI industry capitalizes on, and how climate AI shapes material investment flows and landscapes. Highlighting how data-driven approaches to climate crises reproduce injustices already faced by marginalized communities, our analysis contributes to research on environmental data justice.