GSI Seminar - How everyday images and computer vision are shaping (un)imaginaries of climate change: The case of Google Images

Abstract:

Artificial intelligence, specifically computer vision is an increasingly pervasive technology that seeks to identify the elements of any given image, which can in turn be used to categorise that image. I analyse the impact of computer vision on climate change imaginaries and political engagement. I argue that, rather than opening up new visions of climate futures, the affordances of computer vision are helping to cement an ‘unimaginary’ around “everyday images” (Røstvik et al., 2022) of climate change as technical, depoliticised and dehumanised. I illustrate this argument in three parts by focusing on the web’s most important visual gatekeeper: Google Images. First, I review how Google has moved from a page ranking methodology focused on ‘authority’ to one that foregrounds ‘visual consistency’. Second, I provide evidence of this change by using computer vision itself to illustrate image classification processes and effects, showing how the climate unimaginary is stabilising around dehumanised landscapes and scientific charts. Third, draw on the Science and Technology Studies literature to reflect on the impact of these AI-powered dynamics on political engagement, and whether the negative effects of computer vision machines can be halted or even put to more fruitful use in re-imagining climate futures.