Artificial intelligence has been getting a bad rap in popular culture for decades, from HAL in 2001: A Space Odyssey to Skynet in the Terminator series to the sociopathic robots, so often presented as female, in movies like Ex Machina. But as with so much of entertainment, the actual progress and value of new technologies is lost in sensationalist storytelling, as delightful and well-produced as so many of them are.
Actual artificial intelligence is much more limited, focused, and usually much more positive than what is portrayed. The subset of artificial intelligence known as machine learning is proving to be an essential part of our toolkit for understanding and dealing with one of the most challenging issues we’ve faced as humans: climate change.
I’ve been engaged in aspects of data science and task-oriented artificial intelligence professionally for over a decade. More than a year ago, I started going deep on neural net technologies, concepts, and applications to climate change and cleantech especially, especially with my frequent collaborator, David Clement, who assisted me with many of the conceptual aspects of the report. And now, CleanTechnica’s latest report has dropped: Machine Learning: A Transformative Cleantech & Climate Technology.
Our environment and the climate is incredibly complex, with massive data sets that often make it difficult to find important signals among the noise. Our technological solutions are also complex, with complex information requiring unwavering attention to deliver value. And the solutions are being born into a changing world, one where the developed nations can no longer easily and cheaply outsource unwavering attention and diligence to poorly paid humans on the other side of the planet, as those nations have developed themselves and are rejecting the roles that they once embraced out of necessity.
Machine learning has created much more accurate coastal elevation maps, allowing the fractal coastlines surrounding our inhabited continents and islands to be better assessed for climate risk. It is improving efficiency and safety of our cars and trucks. It’s separating waste streams into high-value and disposable elements with high accuracy. It is improving efficient engagement with energy markets by renewable and storage initiatives. It’s improving the quality of the water we use daily. Machine learning experts have issued a call to arms for machine learning climate action with an organizing structure and initial assessment of where the technologies and techniques are best applied.
This report introduces the basics of machine learning’s key concepts, explores global case studies of its application to research and clean technologies, and outlines areas where it’s been rejected or failed to help understand its limits. In many cases, it also draws a pragmatic perspective on the value of the results, as there are many cases where what is discovered to be possible is not magically made probable.
Researchers considering applying machine learning to their climate or technology projects will find this an invaluable resource to help them understand whether it will add value or detract from their efforts. Entrepreneurs looking to solve market problems that they’ve identified will find it valuable to help them determine whether it should be part of their solution or not, and when and where to apply it. People focused on climate action will find it beneficial to learn of the advances of our understanding and the solutions we are engaging with, finding positive examples to share and key nuances of how to think of them. People looking for collaborators in solving complex climate and environmental challenges will find a fully referenced list of deep experts through this report.
The process of researching this report, reading the literature, talking to deep experts globally, and shaping my thoughts into useful words for others has been a growth experience for me. I’m delighted to share the results with the world. Any remaining errors in the report are mine alone.