International Industrial Ecology Day 2021

Pitfalls of the Widespread Use of Artificial Intelligence for Climate

Artificial intelligence has proven to be a key asset in the assessment and preservation of ecology. At the same time, there have been significant concerns raised regarding whether the widespread utilization of AI technology can actually backfire. Particularly, the notably large energy consumption of AI systems themselves have come under scrutiny; especially with the recent popularity of deep learning (DL) since approximately 2012, high-level computations have raised the overall energy consumption by 300,000 times or more. DL is multilayered ML that demands more computing resources than more “traditional” methods, but is the standard for many applications, such as computer vision. In addition, the big tech companies that are driving much of the innovation in this field often have alarming partnerships with oil and gas companies like Chevron, ExxonMobil, and BP. As a result, the interests of the developers of such “AI for climate” systems may not be in line with how society would prefer them to be. We discuss the downfalls of assuming that AI is inherently good and also propose solutions to the issue. Particularly, energy-efficient processing units are going to be a key asset in the near future to make sure that large-scale AI systems used for climate mitigation and adaptation don’t cause more harm than good. In the private sector, technology companies are already adopting new green pledges and commitments that combat the sentiment that they are playing in the interests of fossil fuel companies.

Author(s)

Name Affiliation
Thomas Chen Academy for Mathematics, Science, and Engineering

« Back to The Future of Artificial Intelligence in the Context of Industrial Ecology: Abstract presentations