GenAI Series - Towards Environmental Responsible GenAI: Carbon, Water and Beyond
2:00 – 3:00 pm MST
This workshop is part of the Generative AI in Scientific Software and Coding Workshop Series (GenAI). You only need to register once to attend the related events in the series.
Abstract
Generative Artificial Intelligence (GenAI) models, while offering solutions to pressing issues like climate change, consume significant energy during training and deployment. This talk will quantify the energy consumption of popular GenAI models, comparing it to other sectors. It will then assess the environmental impact, converting energy consumption into carbon emissions and water usage, and highlight the burden on local communities. The talk will delve into the details of environmental footprint modeling, discussing key factors influencing this conversion. Finally, it will propose sustainable strategies, such as intelligent scheduling, to reduce the environmental footprint of GenAI while ensuring a socially equitable approach. By understanding the energy and environmental implications of GenAI, we can pave the way for a more sustainable future.
Name
Pengfei Li
Pengfei Li is a final-year Ph.D. candidate in Computer Science at the University of California, Riverside, advised by Professor Shaolei Ren. Prior to his Ph.D., Pengfei obtained an M.S.E. degree from the Laboratory for Computational Sensing and Robotics (LCSR) at Johns Hopkins University, advised by Professors Alan Yuille and Gregory Hager.
Pengfei's research aims to develop trustworthy, equitable, and sustainable machine learning systems that improve the efficiency of real-world systems and address critical societal and environmental challenges. Beyond his academic contributions, Pengfei's research has made a substantial impact on public awareness of sustainability and equity issues. His work has been highlighted in over 1000 news outlets worldwide, including prestigious publications like The Associated Press, The Wall Street Journal, Bloomberg, Forbes, and New Scientist. His research findings have also been recognized by international organizations such as UNESCO and the OECD.