Guest essay by Eric Worrall
They don’t actually mention China by name, what they say is “This can be used worldwide in places that aren’t monitoring, and we don’t have to ask permission.”.
How artificial intelligence can tackle climate change
The biggest challenge on the planet might benefit from machine learning to help with solutions. Here are a just a few.
BY JACKIE SNOW
PUBLISHED JULY 18, 2019
Seeing a chance to help the cause, some of the biggest names in AI and machine learning—a discipline within the field—recently published a paper called “Tackling Climate Change with Machine Learning.” The paper, which was discussed at a workshop during a major AI conference in June, was a “call to arms” to bring researchers together, said David Rolnick, a University of Pennsylvania postdoctoral student and one of the authors.
“It’s surprising how many problems machine learning can meaningfully contribute to,” says Rolnick, who also helped organize the June workshop.
AI can also unlock new insights from the massive amounts of complex climate simulations generated by the field of climate modeling, which has come a long way since the first system was created at Princeton in the 1960s. Of the dozens of models that have since come into existence, all look at data regarding atmosphere, oceans, land, cryosphere, or ice.
But, even with agreement on basic scientific assumptions, Claire Monteleoni, a computer science professor at the University of Colorado, Boulder and a co-founder of climate informatics, points out that while the models generally agree in the short term, differences emerge when it comes to long-term forecasts.
“There’s a lot of uncertainty,” Monteleoni said. “They don’t even agree on how precipitation will change in the future.”
One project Monteleoni worked on uses machine learning algorithms to combine the predictions of the approximately 30 climate models used by the Intergovernmental Panel on Climate Change. Better predictions can help officials make informed climate policy, allow governments to prepare for change, and potentially uncover areas that could reverse some effects of climate change.
A grant from Google is expanding the nonprofit’s satellite imagery efforts to include gas-powered plants’ emissions and get a better sense of where air pollution is coming from. While there are continuous monitoring systems near power plants that can measure CO2 emissions more directly, they do not have global reach.
“This can be used worldwide in places that aren’t monitoring,” said Durand D’souza, a data scientist at Carbon Tracker. “And we don’t have to ask permission.”
…Read more: https://www.nationalgeographic.com/environment/2019/07/artificial-intelligence-climate-change/
Nice to see admissions of deep uncertainty and the need for better predictions to make informed climate policy.
But there is a problem the AI experts are kind of glossing over.
I’ve worked on AI before, the most powerful AI I created extracted text from very noisy images, like the following:
What is the big deal? The text is easy for a human to read.
But an AI really struggles to read such images. The reason is, AIs find it very difficult to differentiate between the text, and the color splashes and swirly lines.
To have any chance of the AI successfully reading the text in the image, a lot of human intervention is required; a human has to try to develop a digital filter which cleans as much of the distracting swirly stuff away as possible, leaving an image something like the following:
This incidentally is why websites sometimes give you tests to prove you are not a robot. Most humans can easily pick out the text from a noisy image on a “Are you a robot?” test. AI web robots not so much.
My point is current generation AIs usually do a very poor job of separating noise from data. In the same way a text reading AI can be confused by a few swirly lines, so cutting edge facial recognition AIs can be confused by wearing funny patterned glasses. Tricks which wouldn’t fool a human toddler leave the most powerful current generation AIs floundering.
So I’m really looking forward to seeing the results of AI researcher’s optimistic attempts to use AI to pick useful insights out of the noisy mess which is the world’s climate models.