AI's Environmental Crossroads: Opportunities and Risks

0:00Today on the podcast, we're going to talk about the environmental or some of the environmental impacts of AI. And so this is actually a really interesting topic. It's got a lot of controversy in the past for a number of different reasons. You know, some people say that some people point at the fact that AI, which is very computational heavy, of course, as we all know, uses an incredible amount of electricity. A lot of that is not, you know, clean electricity, quote unquote. Quote, they also point to the fact that AI uses an insane amount of water to train these AI models, something I had never thought about before reading that article, where essentially in order to cool a lot of these AI facilities or these different cloud computing facilities, they use an insane amount of water.

0:47You know, we're talking about lakes worth to really make sure all of the computer hardware works. So there's a lot of these kind of unthought of factors, I think, that go into this. And so today on the podcast, we're going to dive into what NVIDIA's approach is going to be at this. The CEO of NVIDIA was recently at a conference and he talked about this. So today on the podcast, we'll dive into that and talk about this whole field of AI and kind of environmental advancement. So the first thing to say is I really do believe that AI has the ability to accelerate computing, which is quite key to environmental advancements.

1:24And this is also the view that is held by NVIDIA CEO Jensen Huang. And he recently spoke at a summit in Berlin on this topic. So I believe that artificial intelligence and accelerated computing have the potential to drive significant advancements in environmental research. And that is also a view held by Huang. So he made a bunch of remarks that kind of essentially said that during a recent keynote speech at the Berlin Summit for the Earth Virtualization Engines Initiative on Monday or the EVE. And so he was drawing on Richard Frayman's famous quote. He said, what I can't create, I don't understand.

2:06So he emphasized the importance of understanding our environment through modeling and speaking to about 180 people that were at this event, which is a hub for the region's scientific and kind of research community. He really underscored the critical role of their work for policymakers, researchers, and industries. So I think that the summit actually happens. It's a global, it has global participation for this entire summit. But they really leverage AI and high performance computing for improved environmental predictions. And that's what they talk about. So during his speech, he proposed what he says are three miracles, which he says are necessary for researchers to achieve their goals.

2:51And he also highlighted NVIDIA's commitment to these endeavors through its Earth 2 initiative. So the first miracle, according to him, is to simulate environmental conditions rapidly and with a high resolution of just a couple of square kilometers. And then the second involves pre-computing extensive quantities of data. The third miracle, as he says, essentially necessitates visualizing this data interactively with NVIDIA's Omniverse, which is a project they're currently working on. So this EVE initiative and the next wave of, you know, environmental innovation is essentially an international collaboration providing really easy, accessible kilometer scale environmental information for sustainable planet management.

3:43And it has been built on considerable advancements over the last, I would say, 25 years. And this is working directly with NVIDIA's Earth 2 project. So EVE aims to speed up advancements in high-resolution environmental projections. And also accelerated computing is, I believe, already really pushing a number of different applications, including ICON, IFS, NEMO, MPAS, WRFG, and a bunch of other ones. But essentially, more computing power is coming. And with the introduction of NVIDIA's GH200 Gracehopper Superchip, I think that this cutting-edge accelerated CPU, which is designed for large-scale AI and HPC applications, it delivers up to 10 times higher performance for applications handling terabytes of data.

4:39So this is a really massive improvement in technology, which is actually allowing a lot of this new environmental tech to come forward. Because, you know, as we know, when we're trying to make models and predictions of the environment and impacts that different things are having on the environment, it is taking an insane amount of resources. There's so many different factors to look at when it comes to environmental changes and when it comes to, you know, really preserving the environment we have. And so I think this is a really, really impressive initiative that NVIDIA is, you know, attempting to undertake.

5:17So Huang went on to highlight at this keynote that he gave NVIDIA's Modulus, which essentially is an open source framework for building, training, and fine-tuning physics-based machine learning models. And also ForecastNet, which is a data-driven forecasting model. So using raw data alone, ForecastNet can learn and learn essentially the principles behind complex environmental patterns, which are demonstrated by actively predicting the path of Hurricane Harvey through modeling the Coriolis force. Right. So this is absolutely insane. Right. Like we're creating technology that can actually predict the path of hurricanes and all sorts of other things that have big impacts on the environment.

6:01So he illustrated how NVIDIA's technologies promised to make essentially this immense pool of knowledge more accessible. He presented a high resolution interactive visualization of global scale environmental data in the cloud using digital twins to create complex models of intricate systems. and as he kind of wrapped up this this keynote speech that he gave he thanked a lot of the researchers there at the event and suggested um a playful mission for eve and uh he's referencing star trek the star trek phrase of course and he described the earth as the final frontier with eve's mission to push the boundaries of computing in service of understanding and preserving our environment.

6:46So I think as AI and accelerated computing continue to develop, it's clear that tech leaders over at NVIDIA and a lot of other places see them as a really crucial tool in the quest to understand and protect our environment, making a new era of environmental innovation. And I, for one, think this is really impressive. A lot of the steps that are being taken, just the sheer size of these models is quite impressive. And it's going to be a very interesting area to continue to follow and see what advancements are made in the future.