Tackling climate change with machine learning.
Rolnick, David, et al. “Tackling climate change with machine learning.” arXiv preprint arXiv:1906.05433 (2019).
This paper was quite a marathon but it was entirely worth the effort, being one of the most broadly educative papers I’ve read recently. Climate change is such a huge topic that I am not even exaggerating to say that it has ties to every corner of science, engineering and industry I can think of.
This means that there is an almost endless number of levers we can pull on to affect our climate future - which makes approaching the problem quite overwhelming. On the bright side, many of the opportunities here are valuable in their own right, so I have a sense that forward-looking industries are already working in the right direction (although we do need to have our priorities right). Importantly, there are large amounts of overlap between many of the levers. Fundamental progress in any of the key capabilities listed on the last slide could lead to important impact across many different challenges, and this is indeed why tools from machine learning can have a lot of leverage.
That said, as the authors do note, technology is really only one part of the problem. Across every chapter of the paper, there are many points on the importance of generating data and building models to inform policy-making, which must be fully appreciated. At the end of the day, most global-scale problems are really not going to be solved by some “hot new tech” to save the world, but via better-informed policies that provide scaffolding for humans to behave better. There is no easy way out.
Call it a can of worms, Pandora’s box, or a rabbit hole - either way, my eyes have been opened to a new universe of interesting ideas and challenges, and an armful of new papers have been added to my ever-growing queue.
View in Google Slides.
Join the conversation on:
