Computing, and AI in particular, hold substantial potential to both help and hinder climate change mitigation. They have been called a major threat – responsible for large and growing emissions – as well as a critical enabler of emission reductions across other sectors. Reflecting the use of inconsistent and often problematic methodologies, estimates of the sector’s footprint diverge by an order of magnitude, frustrating public understanding and effective policymaking. Meanwhile, its sustainability benefits are uncertain and its detrimental indirect consequences (such as enabling more efficiency in the oil & gas sector) often overlooked.
Set in the wider context of computing, this short course will provide an overview of the complex relation between AI and sustainability. It addresses AI’s footprint (energy, GHGs, water, resources, and pollution) as well as its far less understood – but likely more important – levers to bring about environmentally beneficial or detrimental consequences across society. It will particularly focus on the methodological challenges and possible solutions, hoping to spark the participants’ interest for this quintessentially transdisciplinary field.