Veranstaltung auf Englisch.

Recent developments in artificial intelligence (AI) have led to breakthroughs across scientific discovery, language understanding, and generative modelling. This progress has been driven by enormous amounts of data and massive computing power, which carry a growing cost for the environment and for the people whose data and labour AI systems are built on. Data has now become a bottleneck for continued scaling, shifting the field from building models from data to actively creating data for building models – a trend characterised as hyper-datafication. A concrete example of this trend is the collection of egocentric data – that is, video recorded from a first-person perspective. This video captures everyday life and jobs and is collected to build robotic AI systems capable of completing human tasks.  The talk gives concrete examples of hyper-datafication, its economic, environmental, and social costs, and concrete proposals for reducing them.

Sophia N. Wilson is a Ph.D. fellow at the Sustainable Artificial Intelligence for Sciences (SAINTS) Lab, Department of Computer Science, University of Copenhagen, supervised by Raghavendra Selvan and Erik Dam. Her research spans two strands: a technical strand developing algorithms and AI systems that are more resource efficient in data, compute, and energy use; and a socio-technical strand measuring and mitigating sustainability harms in the AI industry, including environmental impact and poor labour conditions.

Organisation: Prof. Dr. Jan Distelmeyer (Fachhochschule Potsdam/Universität Potsdam)

Wir bitten um vorherige Anmeldung unter .