Abstract:
In his talk, Erik Ljungberg will speak about the infrastructural conditions that preceded the adoption of machine learning in the field of Swedish forestry. It is guided by the notion that in processes of knowledge production one must attend simultaneously to cultural, technological and epistemic factors.

By analyzing policy documents, forestry journals and interviews with key actors in the contemporary Swedish forest research community, he traces three pivotal developments that have shaped the current regime of forest monitoring in Sweden. The first is the advent of new technologies for data collection, such as laser scanners, drones, and high-resolution satellite imagery, which gradually superseded traditional methods such as aerial photography and field sampling starting in 2003. Second, the integration of statistical models from the field of data science, ranging from linear regression to neural nets and random forests, which began to complement field-based measurements in the early 2000s. And third is the progressive onset of environmental policy objectives beginning in 1997 which aimed to give equal weight to productivity and sustainability.

These developments have set the stage for a potential reorganization of the forestry sector, where newfound digital methods represent increased monitoring capacities. Yet, as some seek to use these methods for increased productivity, while others view them as tools for sustainability, a dynamic of friction has emerged. Finally, he discusses whether the transition from field samples and aerial photography to data points within probability distributions represents a epistemic shift in how information about the forest is perceived and understood.

Bio:
Erik Ljungberg has a master's degree in history of knowledge and a second bachelors in cultural anthropology from the University of Oslo. He is currently a doctoral student the Division of History of Science, Technology and Environment at the Royal Institute of Technology in Stockholm. Currently he is part the project "The Mediated Planet" which looks the global environment as emerging through practices of collecting environmental data. His current research focuses the digitization of Swedish forestry between 1980 and today with a special focus on the importance of machine learning and artificial intelligence. His disciplinary homes are located in science and technology studies, media theory and the environmental humanities.

Coffee, tea, cake and fruit will be served before the colloquium @ 2 pm.