Dr. Katie Detwiler, Assistant Teaching Professor of Science and Technology Studies at Stevens Institute of Technology, will speak about the environmental politics of AI, which are often framed around data centers' energy and water use, and with good reason. She argues that the extractive dimensions of AI run even deeper and further back—into practices of dataset production in domains that seem far afield from contemporary AI and its usual sources of training data, and in places that seem far afield from metropolitan imaginaries of AI innovation. Drawing on ethnographic fieldwork among astronomers, engineers, and miners in Chile's Atacama Desert, Dr. Detwiler will trace how contemporary data practices in the seemingly disparate fields of astronomical observation and industrial mineral extraction enact an environmental politics that is prior to, but constitutive of, AI's downstream ecological effects.
