WEBVTT - NASA and AI: Decoding Our Universe 0:00:00.120 --> 0:00:04.240 Welcome to Tech Stuff, a production from iHeartRadio. This season 0:00:04.320 --> 0:00:07.880 on Smart Talks with IBM, Malcolm Glabwell is back, and 0:00:07.920 --> 0:00:10.600 this time he's taking the show on the road. Malcolm 0:00:10.640 --> 0:00:14.720 is stepping outside the studio to explore how IBM clients 0:00:14.760 --> 0:00:18.759 are using artificial intelligence to solve real world challenges and 0:00:18.840 --> 0:00:23.000 transform the way they do business, from accelerating scientific breakthroughs 0:00:23.079 --> 0:00:27.520 to reimagining education. It's a fresh look at innovation in action, 0:00:27.960 --> 0:00:31.720 where big ideas meet cutting edge solutions. You'll hear from 0:00:31.720 --> 0:00:36.000 industry leaders, creative thinkers, and of course Malcolm Glabwell himself 0:00:36.240 --> 0:00:39.879 as he guides you through each story. New episodes of 0:00:39.920 --> 0:00:43.280 Smart Talks with IBM drop every month on the iHeartRadio app, 0:00:43.440 --> 0:00:47.240 Apple Podcasts, or wherever you get your podcasts. Learn more 0:00:47.280 --> 0:00:49.920 at IBM dot com slash smart Talks. 0:00:52.520 --> 0:00:55.440 If I were to go back, I don't know, thirty 0:00:55.520 --> 0:00:59.600 years in Kenya, what's the difference between then and now? 0:00:59.640 --> 0:01:04.360 Intern of tree Cover, I'm talking to Philip, thego Special 0:01:04.400 --> 0:01:06.640 Technology envoid to the Kenyan President. 0:01:07.240 --> 0:01:10.160 Let's speak as if you think about we arena elevent 0:01:10.200 --> 0:01:13.200 trol perscs and previously we were more than twenty percent. 0:01:13.600 --> 0:01:17.000 So we are cutting trees more than we're planting them. 0:01:17.280 --> 0:01:21.000 In thirty years, Kenya lost half its tree cover half. 0:01:21.720 --> 0:01:25.200 And here's why that matters. Kenya is a mountainous country. 0:01:25.720 --> 0:01:28.560 Dotted throughout the highlands are dozens of what canyons call 0:01:28.720 --> 0:01:34.919 water towers, natural reservoirs, densely forested areas capable of absorbing 0:01:34.959 --> 0:01:37.200 the enormous amount of water that falls on the country 0:01:37.280 --> 0:01:41.880 during the rainy seasons. The tree roots and undergrowth secure 0:01:42.000 --> 0:01:45.640 and capture moisture, then slowly release it into the rivers 0:01:45.640 --> 0:01:49.000 that flow down into the country's low lying coastal areas. 0:01:49.640 --> 0:01:54.520 But in recent years the water towers have depleted, settlements 0:01:54.680 --> 0:01:58.480 have encroached on them, trees have been chopped down, thousands 0:01:58.560 --> 0:02:02.280 of acres cleared, the natural reservoirs cease to hold nearly 0:02:02.280 --> 0:02:05.800 as much water, so now Kenya is prone to extremes. 0:02:06.240 --> 0:02:08.760 Too much water flowing down from the highlands in the 0:02:08.840 --> 0:02:12.360 rainy season and too little water left during the dry season. 0:02:13.080 --> 0:02:15.120 So you have a couple of hours of water, then 0:02:15.120 --> 0:02:17.440 you have a couple of hours with no water that 0:02:17.639 --> 0:02:20.560 taps off to be dry by the city authority. So 0:02:20.639 --> 0:02:23.840 that's the significance of the water towers. We have when 0:02:23.880 --> 0:02:24.800 they cannot hold water. 0:02:25.240 --> 0:02:29.440 Kenya desperately needed to restore its water towers by planting 0:02:29.760 --> 0:02:33.239 as many trees as humanly possible. So in the fall 0:02:33.280 --> 0:02:37.120 of twenty twenty three, the Kenyan government took action. It 0:02:37.240 --> 0:02:41.799 started a national holiday, National Tree Growing Day, a day 0:02:41.840 --> 0:02:44.560 to allow the citizens of Kenya to go out into 0:02:44.600 --> 0:02:47.919 the forest to dominate the Kenyan countryside and plant as 0:02:48.000 --> 0:02:52.080 many trees as they can, and the government decided on 0:02:52.120 --> 0:02:52.640 a number. 0:02:53.520 --> 0:02:58.480 The presidents really focus right around how to ensure that 0:02:58.520 --> 0:03:00.880 we don't lose more for us was in this very 0:03:00.919 --> 0:03:03.240 ambitious campaign around fifteen billion trees. 0:03:03.760 --> 0:03:07.280 That's right, fifteen billion with a bee. 0:03:07.840 --> 0:03:10.120 So imagine that number will tell you the ambition, not 0:03:10.240 --> 0:03:12.840 as he tells you the deficit. It has to be 0:03:12.880 --> 0:03:15.160 fifteen billion in the next eight years. 0:03:15.520 --> 0:03:18.799 Fifteen billion trees over eight years averages out to more 0:03:18.800 --> 0:03:22.880 than five million trees per day. That's a lot of trees. 0:03:23.440 --> 0:03:26.040 But with such a massive goal, how can you track 0:03:26.080 --> 0:03:28.760 your progress? How do you know where to plant those 0:03:28.800 --> 0:03:31.920 trees so they'll have the most impact. How do you 0:03:32.040 --> 0:03:35.920 monitor where older trees are still being cut down? Well, 0:03:35.960 --> 0:03:39.640 the answer to those questions came from IBM and a 0:03:39.680 --> 0:03:45.560 little space agency called NASA. That's right, folks, Smart Talks 0:03:46.280 --> 0:03:50.400 is going to space. My name is Malcolm Glawell. You're 0:03:50.400 --> 0:03:53.920 listening to the latest episode of Smart Talks with IBM, 0:03:54.280 --> 0:03:57.040 where we offer our listeners a glimpse behind the curtain 0:03:57.160 --> 0:04:00.960 of the world of technology. In this season, IBM has 0:04:01.000 --> 0:04:05.880 gone inside elementary school classrooms, toured formulation labs at Loreel, 0:04:06.200 --> 0:04:10.080 and spoken with the fan development team at Scuderia Ferrari HP. 0:04:11.080 --> 0:04:15.600 In this episode, how IBM is partnering with NASA to 0:04:15.680 --> 0:04:20.200 build geospatial models using data from satellites to better understand 0:04:20.200 --> 0:04:26.080 our Earth and Solar system. 0:04:26.160 --> 0:04:34.560 Five four three two one zero all engine running liptoff. 0:04:34.600 --> 0:04:37.719 We have a liptoff thirty two minutes past the hour 0:04:38.080 --> 0:04:39.440 liftoff on Apollo eleven. 0:04:40.120 --> 0:04:43.800 IBM has worked on space related projects since before I 0:04:43.839 --> 0:04:48.400 was even born. Im all for Man, A team of 0:04:48.440 --> 0:04:52.320 four thousand IBM engineers helped create the Saturn five rocket 0:04:52.320 --> 0:04:58.640 that took Neil Armstrong to the Moon. Arm And when 0:04:58.640 --> 0:05:01.640 I think of NASA, I tend to picture the moon landing, 0:05:02.160 --> 0:05:04.560 or the team of people back in Houston guiding the 0:05:04.600 --> 0:05:08.520 Apollo mission, or the Hubble telescope or astronauts aboard the 0:05:08.560 --> 0:05:13.080 International Space Station. What I didn't think about until now 0:05:13.600 --> 0:05:15.320 are NASA's geographers. 0:05:16.120 --> 0:05:18.039 In order to go places, you need at map things. 0:05:18.520 --> 0:05:22.560 This is Kevin Murphy, chief Science Data Officer at NASA's 0:05:22.680 --> 0:05:24.039 Science Mission Directorate. 0:05:24.760 --> 0:05:27.760 But I think that there's an assumption that NASAs all 0:05:27.800 --> 0:05:31.560 about rockets and astronauts, and certainly that's a really large 0:05:31.600 --> 0:05:32.920 part and important part in NASA. 0:05:33.720 --> 0:05:37.080 NASA sends people to space and looks out of the stars, 0:05:37.560 --> 0:05:41.160 but NASA also looks down at the Earth. The agency 0:05:41.160 --> 0:05:46.960 has about one hundred and fifty satellites that use radar, lightar, landset, aquatera, 0:05:47.120 --> 0:05:52.599 cloudset AURA, low Earth orbit, medium Earth orbit, geostationary orbit, 0:05:52.800 --> 0:05:57.200 on and on. In one sense, NASA makes hardware. They 0:05:57.200 --> 0:06:00.800 build rockets and spacecraft and all those delights that circle 0:06:00.839 --> 0:06:06.720 the Earth. But fundamentally NASA also collects data. It's scientists 0:06:06.760 --> 0:06:09.600 and the engineers people like Kevin want to make the 0:06:09.640 --> 0:06:13.560 best use possible of all the information gathered by all 0:06:13.680 --> 0:06:15.800 those many dozens of instruments. 0:06:16.360 --> 0:06:20.080 Right now, we gather around twenty five petabytes of new 0:06:20.240 --> 0:06:23.200 observational data per year. In the next couple of months, 0:06:23.200 --> 0:06:28.720 we're about to launch a high resolution Global Radar when 0:06:28.760 --> 0:06:32.320 that launches, will double how much we collect every year 0:06:32.400 --> 0:06:34.440 to about fifty petabytes of information. 0:06:35.120 --> 0:06:39.160 Actually, since we recorded this conversation, NASA launched that global 0:06:39.240 --> 0:06:43.920 radar what they call NYSAR. So NASA is already generating 0:06:43.960 --> 0:06:46.880 new data at the rate of fifty petabytes each year. 0:06:47.520 --> 0:06:50.800 To put that in perspective, a single petabyte could hold 0:06:50.839 --> 0:06:55.560 about five hundred billion pages of standard printed text. Now 0:06:55.600 --> 0:06:58.480 can anyone sort of apply to use this data is. 0:06:58.760 --> 0:07:01.480 They don't even have to apply. It's free and open data. 0:07:01.600 --> 0:07:05.400 It advances how we understand what we do on Earth 0:07:05.480 --> 0:07:09.280 and how we see ourselves within the universe. People can 0:07:09.360 --> 0:07:12.120 take it for so many different downstream applications. So you 0:07:12.160 --> 0:07:15.000 can go to our websites today, you can search through 0:07:15.040 --> 0:07:19.480 our tools, and you can download information from the Mars rovers, 0:07:19.480 --> 0:07:23.160 you can download information from the Lunar Reconnaissance Orbiter or 0:07:23.200 --> 0:07:24.920 any of the Earth Science Data satellites. 0:07:25.200 --> 0:07:28.560 And give me an example of a really cool application, 0:07:29.320 --> 0:07:31.560 a really cool use that someone I don't know in 0:07:31.640 --> 0:07:34.560 academic or whatever has used your data for. It is there? 0:07:34.560 --> 0:07:37.800 It okay? So one of the really kind of cool 0:07:37.920 --> 0:07:41.760 but unexpected observations that we had is that we launched 0:07:41.800 --> 0:07:46.160 a pair of satellites in their early two thousands called Grace, 0:07:46.560 --> 0:07:49.440 and these satellites orbit the Earth and they can measure 0:07:49.560 --> 0:07:52.280 very precisely the distance that they're away from each other 0:07:52.280 --> 0:07:55.000 as they orbit the Earth. And as you go into 0:07:55.000 --> 0:07:58.440 gravity wells, you can actually see a satellite accelerate and 0:07:58.480 --> 0:08:02.240 the other one accelerate after it, right, And using that information, 0:08:02.600 --> 0:08:05.280 we were trying to map kind of the gravity fields 0:08:05.600 --> 0:08:07.920 of Earth. What what they found is that they can 0:08:07.920 --> 0:08:11.360 actually map below kind of the mass of Earth to 0:08:11.800 --> 0:08:15.320 where water storage is. For instance, so aquifers, right, so 0:08:15.600 --> 0:08:20.320 you can monitor through gravity how much water is being 0:08:20.360 --> 0:08:24.320 depleted or added to an aquifer or the density of glaciers. 0:08:24.880 --> 0:08:28.240 So, just to back up for a moment, the presence 0:08:28.400 --> 0:08:32.720 and density of water deposits below the Earth's surface have 0:08:32.800 --> 0:08:38.079 an effect on gravitational fields that are being measured in space. 0:08:38.120 --> 0:08:38.480 Correct. 0:08:39.040 --> 0:08:42.559 Yeah, And so does that tell you presuming you learn 0:08:42.679 --> 0:08:44.600 things like where there's an aquifer where you didn't think 0:08:44.600 --> 0:08:46.040 there was an aquifer. 0:08:46.080 --> 0:08:48.920 Or if it's being depleted faster? Yeah? 0:08:49.080 --> 0:08:51.480 Yeah. So who's using that kind. 0:08:51.360 --> 0:08:55.000 Of data All sorts of different organizations, whether they're you know, 0:08:55.280 --> 0:08:59.120 NGOs or government agencies or people that are planning a 0:08:59.160 --> 0:09:00.439 large agricultural product. 0:09:00.480 --> 0:09:02.600 How did you Was that an intentional desis? 0:09:02.640 --> 0:09:03.959 It wasn't It was accidental. 0:09:04.679 --> 0:09:11.240 It was accidental. NASA has assembled a historically unprecedented mountain 0:09:11.240 --> 0:09:15.079 of data about the physical world, free and open to anyone, 0:09:15.480 --> 0:09:18.240 and the possibilities for how that information can be used 0:09:18.320 --> 0:09:24.120 are so vast that even NASA is still uncovering them. 0:09:24.520 --> 0:09:27.200 When I was a kid, I loved legos. I had 0:09:27.240 --> 0:09:30.320 a huge bin full of them. At the time, legos 0:09:30.400 --> 0:09:33.920 were really just colored bricks of various sizes. They weren't 0:09:33.960 --> 0:09:36.880 as complicated as they are today. And what I realized 0:09:36.920 --> 0:09:39.480 even then was that there were more possibilities in a 0:09:39.520 --> 0:09:42.400 box of Legos than I could ever imagine on my own. 0:09:43.240 --> 0:09:44.880 I played with my brother and he would show me 0:09:44.920 --> 0:09:47.200 something that hadn't occurred to me. And I go to 0:09:47.200 --> 0:09:49.520 my friend Bruce's and see that he was off on 0:09:49.559 --> 0:09:53.040 some legos tangent that I'd never even thought of, like 0:09:53.080 --> 0:09:55.840 a cool bridge or a castle or a truck. I 0:09:56.000 --> 0:09:59.480 used legos one way, Bruce used his legos in a 0:09:59.559 --> 0:10:05.080 completely different way. NASA's data treasure Trove is like a 0:10:05.240 --> 0:10:08.520 very very big box of legos, And here's the question. 0:10:09.080 --> 0:10:14.240 With so much data, containing so many possible connections, could IBM, 0:10:14.520 --> 0:10:20.400 and specifically IBM's artificial intelligence help NASA scientists uncover patterns 0:10:20.400 --> 0:10:23.840 and connect systems in a way they've never done before. 0:10:26.320 --> 0:10:28.800 Everything started with a question, right. 0:10:28.840 --> 0:10:32.960 I'm talking to one Bernabe Moreno, director of IBM Research 0:10:33.000 --> 0:10:33.480 in Europe. 0:10:34.280 --> 0:10:39.000 As we advance AI, we have new tools to understand this, 0:10:39.160 --> 0:10:42.600 around this, understand the world, understand the language, and understand 0:10:42.600 --> 0:10:45.720 our planets. And the question that we were asking ourselves 0:10:45.840 --> 0:10:48.360 was all these new advances that we see in language. 0:10:48.400 --> 0:10:51.720 It was a post GPT moment. Could we apply the 0:10:51.800 --> 0:10:55.320 same idea and the same architecture and technology to a 0:10:55.480 --> 0:10:56.560 data about our planet. 0:10:57.200 --> 0:11:00.560 The advent of AI created a new opportunity. What if 0:11:00.559 --> 0:11:04.160 all of NASA's mountain of data could be organized, analyzed, 0:11:04.440 --> 0:11:09.760 understood by artificial intelligence. The original idea was to create 0:11:09.800 --> 0:11:13.440 a geospatial foundation model for the Earth, and from there 0:11:13.920 --> 0:11:18.640 create additional specialized models for other scientific priorities of NASA, 0:11:19.360 --> 0:11:23.079 and finally quit an AI system that can understand all 0:11:23.160 --> 0:11:26.920 the data across those specialized models in order to uncover 0:11:27.040 --> 0:11:32.240 hidden insights and relationships. Together, these models could unlock an 0:11:32.360 --> 0:11:36.960 infinite number of potential applications. I asked Kevin Murphy at 0:11:37.040 --> 0:11:40.280 NASA about the beginning of these Earth models. 0:11:40.880 --> 0:11:43.559 Has some colleagues, and we were investigating in a number 0:11:43.559 --> 0:11:48.560 of different avenues of using AI with our data, but 0:11:48.679 --> 0:11:51.720 also kind of the management and stewardship of the data, 0:11:51.760 --> 0:11:53.960 so not only like the observations, but how we make 0:11:54.000 --> 0:11:57.760 it available to people, make it discoverable. And they said, hey, 0:11:58.440 --> 0:12:01.000 we see these transform architecture. We think that they can 0:12:01.040 --> 0:12:05.199 be applicable to some of the sequential observations that we make. 0:12:05.760 --> 0:12:07.840 We'd really like to work with IBM on that. And 0:12:07.880 --> 0:12:12.120 I was like, I'm really skeptical, but because I hadn't 0:12:12.200 --> 0:12:18.240 seen those types of tools really produce results that were 0:12:18.960 --> 0:12:22.000 commensurate with the amount of effort you put into them, right, 0:12:22.040 --> 0:12:24.360 So we were getting some really good results and deep 0:12:24.480 --> 0:12:27.720 learning approaches, but they took a lot of effort. 0:12:28.000 --> 0:12:29.600 But Kevin came around quickly. 0:12:30.440 --> 0:12:35.120 When we typically develop a new data product or an algorithm, 0:12:35.600 --> 0:12:39.320 it takes anywhere from you know, twelve months, eighteen months, 0:12:39.320 --> 0:12:44.719 twenty four months to go from data and hypothesis to 0:12:44.960 --> 0:12:48.880 results which is validated. We were able to get approximately 0:12:48.920 --> 0:12:54.000 the same precision for some well known types of benchmarks 0:12:54.400 --> 0:12:56.800 with and I think it was about four months of 0:12:56.880 --> 0:12:57.560 starting the work. 0:12:57.679 --> 0:13:01.840 Yeah, yeah, so it happened faster than you thought, much faster. 0:13:02.920 --> 0:13:06.560 In twenty twenty three, IBM and NASA launched a foundation 0:13:06.760 --> 0:13:11.560 model trained on NASA's harmonized landset sentinel to satellite data 0:13:11.960 --> 0:13:15.960 across the continental United States. They named the model Prithvi, 0:13:16.480 --> 0:13:20.439 the Sanskrit word for Earth. The first version of Prithvi 0:13:20.920 --> 0:13:24.960 used only Earth observation images and just that was enough 0:13:25.000 --> 0:13:29.000 to totally change Kevin's idea of what foundation models could do. 0:13:30.040 --> 0:13:33.920 But they didn't stop there. IBM and NASA were encouraged 0:13:34.080 --> 0:13:38.400 at how well Prithvy worked for Earth observation tasks, so 0:13:38.720 --> 0:13:41.520 they decided to create a more complex version of Prithvy 0:13:41.920 --> 0:13:46.040 that could understand whether and climate data. They hoped this 0:13:46.080 --> 0:13:49.200 new version of Prithvi would allow researchers to answer new 0:13:49.280 --> 0:13:52.920 questions about the Earth, from short term weather forecasting to 0:13:53.000 --> 0:13:56.640 longer term climate effects. Imagine you have a map of 0:13:56.800 --> 0:14:01.800 all the different temperatures, pressures, clouds, rainfall and more from 0:14:01.840 --> 0:14:06.120 around the globe. With this map, IBM and NASA could 0:14:06.160 --> 0:14:10.560 implement advanced tasks. They could track the formation of al Nino, 0:14:11.040 --> 0:14:13.280 or predict how the path of a hurricane would change 0:14:13.600 --> 0:14:17.120 if the ocean temperature went up by half a degree. 0:14:17.400 --> 0:14:20.520 I would always remember this moment was when we created 0:14:20.520 --> 0:14:25.640 the Weather and Climate Foundational Model. The senior methodologist of NASA, 0:14:26.080 --> 0:14:29.200 it was like, I cannot believe that it has changed the 0:14:29.280 --> 0:14:31.440 way I think about the AI and ever since, he's 0:14:31.480 --> 0:14:34.000 been kind of preaching with this A sample. 0:14:34.400 --> 0:14:36.880 One and his team then took the model and decided 0:14:36.920 --> 0:14:41.000 to test it, really tested it. Took away ninety nine 0:14:41.040 --> 0:14:44.320 percent of the data points and ran the experiment again. 0:14:44.960 --> 0:14:47.040 What they were trying to figure out is if the 0:14:47.080 --> 0:14:50.120 model had learned enough about the basic principles of the Earth, 0:14:50.520 --> 0:14:53.480 the underlying physics of the way the planet works, to 0:14:53.600 --> 0:14:56.920 fill in the blanks on its own with just one 0:14:57.000 --> 0:15:00.600 percent of the original data, would it still be accurate 0:15:00.640 --> 0:15:07.200 in its predictions. What happened. The model crushed it so 0:15:07.440 --> 0:15:09.840 it was able to extrapolate on the basis of one 0:15:09.920 --> 0:15:13.320 percent of the data what the entire picture looked like. 0:15:13.440 --> 0:15:17.480 Yes, because pre learned everything right. 0:15:17.400 --> 0:15:20.480 Yeah, it learned the kind of principles of exactly Yeah. 0:15:20.840 --> 0:15:24.280 Oh wow, that's very very impressive. So at that moment 0:15:24.760 --> 0:15:28.840 when you realize you could do that and just curious 0:15:28.840 --> 0:15:31.720 about your emotional I mean, did you jump up and down? 0:15:31.760 --> 0:15:32.280 What did you do? 0:15:32.760 --> 0:15:36.200 So it's like, wow, it was a very emotional meeting 0:15:36.240 --> 0:15:41.600 because you know, having this person say now I'm convinced right, Yeah, 0:15:41.720 --> 0:15:44.600 it was kind of a quite a special moment. These 0:15:44.600 --> 0:15:46.280 moments make your life as a researcher. 0:15:47.720 --> 0:15:51.400 Ibm And as a launch prithe for Weather and Climate 0:15:51.480 --> 0:15:54.520 in twenty twenty four and while ibm And as a 0:15:54.520 --> 0:15:58.520 scientist could use Prithvy to run interesting experiments, they were 0:15:58.520 --> 0:16:02.080 even more excited about how Prithy could help people in 0:16:02.120 --> 0:16:10.400 the real world. So let's go back to Kenya Ambassador 0:16:10.480 --> 0:16:14.080 Philip Diego and the country's great tree planting project. 0:16:14.920 --> 0:16:17.880 So on those initial months, there was a massive effort, 0:16:17.960 --> 0:16:23.080 including a couple of national holidays for tree planting. Yes, 0:16:24.000 --> 0:16:26.200 where the entire cabinet was sent. 0:16:26.520 --> 0:16:28.560 Ah, did you plant trees. 0:16:28.440 --> 0:16:29.080 As I did? 0:16:29.160 --> 0:16:31.040 Oh my god, I said, the entire cabinet plus someone 0:16:31.240 --> 0:16:32.040 we have to be seen. 0:16:32.480 --> 0:16:34.280 Are you good at the planet? Two weeks ago? 0:16:34.640 --> 0:16:36.520 Well, it's very easy to go hole put a tree 0:16:36.600 --> 0:16:38.720 in the ground. 0:16:38.240 --> 0:16:42.040 Well wow, what planting a tree is easy? But remember 0:16:42.480 --> 0:16:47.400 it has to happen fifteen billion times. IBM research has 0:16:47.440 --> 0:16:51.200 been operating in Nairobi since twenty thirteen, and what ken 0:16:51.240 --> 0:16:55.200 you wanted, at least in the beginning was straightforward. The 0:16:55.280 --> 0:16:58.280 prith Fee model that IBM and NASA built could be 0:16:58.400 --> 0:17:02.480 used to essentially make the world old's greatest map, and Kenya, 0:17:02.600 --> 0:17:05.480 with IBM's help, could use that model to make the 0:17:05.520 --> 0:17:09.440 world's greatest map of Kenya. The first step was to 0:17:09.520 --> 0:17:12.560 lay a grid across a topography of the country, break 0:17:12.600 --> 0:17:16.480 the forest into manageable bite sized pieces, each of which 0:17:16.480 --> 0:17:18.000 could be analyzed separately. 0:17:18.840 --> 0:17:21.119 So because our forest is massive when you look at 0:17:21.160 --> 0:17:23.800 it in terms of green hite, but only lay it, 0:17:24.040 --> 0:17:26.720 you're able to break it into pieces, like into boxes. 0:17:26.920 --> 0:17:30.560 And for us that was important because then it's easy 0:17:30.640 --> 0:17:33.400 to tackle it when it's in a greed system than 0:17:33.560 --> 0:17:36.040 just as a massive forest. So that was also what 0:17:36.720 --> 0:17:38.120 the model was able to do. 0:17:38.560 --> 0:17:42.119 Then the model painstakingly sorted through each of those boxes 0:17:42.520 --> 0:17:45.720 and look for what Philip calls hotspots, so. 0:17:45.640 --> 0:17:48.320 You can see, for example, very quickly which other areas 0:17:48.320 --> 0:17:51.520 are being eroded very fast, and that you need to 0:17:51.600 --> 0:17:54.800 quickly protect. Yeah, because you sometimes and that's where you 0:17:54.800 --> 0:17:56.600 want to target, right, I mean it's not possible to 0:17:56.600 --> 0:17:58.240 do everything at the same time. 0:17:58.400 --> 0:18:00.199 Do you have a definition of a hotspot and how 0:18:00.240 --> 0:18:02.440 many hotspots are there according to that definition? 0:18:03.320 --> 0:18:03.920 Oh, there are a lot. 0:18:04.040 --> 0:18:07.320 So we have more than forty water towers, and I'll 0:18:07.320 --> 0:18:10.320 tell you all of them have hotspots. And the hot 0:18:10.320 --> 0:18:14.479 spots in my definition areas that are being degraded faster 0:18:14.640 --> 0:18:17.600 and in a very unusual way. Right, you can literally 0:18:17.640 --> 0:18:21.040 see how human activity is seriously degrading that particular area 0:18:21.400 --> 0:18:23.600 that if you do not have a direct intervention, we'll 0:18:23.640 --> 0:18:27.000 lose the entire forest. So that's the hotspot for us, 0:18:27.560 --> 0:18:29.520 because you think about cutting one hundred trees a day 0:18:29.600 --> 0:18:31.560 and cutting a million trees a day, So that's a 0:18:31.640 --> 0:18:34.680 hot spot. You want to look at places where there's 0:18:34.920 --> 0:18:39.600 just unusually high activity of deforestation in a hotspot. 0:18:39.880 --> 0:18:42.120 The size of each box in the grid was ten 0:18:42.160 --> 0:18:45.880 by ten meters, about half a tennis court. That's how 0:18:46.000 --> 0:18:50.119 closely they were examining the forest, so very crudely. The 0:18:50.160 --> 0:18:55.240 model ingests all of this satellite data and it helps 0:18:55.240 --> 0:18:58.560 you answer some very specific questions like where should we 0:18:58.600 --> 0:19:03.800 prioritize our tree planning efforts which areas down to an 0:19:03.800 --> 0:19:08.639 extraordinary level of specificity are eroding most quickly. You know, 0:19:08.680 --> 0:19:11.879 all those kinds of practical questions about how to direct 0:19:11.920 --> 0:19:12.600 your strategy. 0:19:12.920 --> 0:19:15.000 So if you think about a smart forest, right, and 0:19:15.040 --> 0:19:17.679 that's really for us, we're calling it smart fencing, smart forests, 0:19:17.680 --> 0:19:21.400 everything that's smart because of AI. If you think about 0:19:22.119 --> 0:19:24.800 your usual what you can see with your eyes and 0:19:24.840 --> 0:19:27.879 then the satellite layer which just zooms in and you 0:19:27.920 --> 0:19:30.760 see green. So what the model has been able to 0:19:30.760 --> 0:19:33.080 do is to create a smart layer, right, and then 0:19:33.200 --> 0:19:37.440 that smart layer you can actually see many things, from analytics, 0:19:37.440 --> 0:19:40.320 to the greeds, to a dashboard, one a lot. So 0:19:40.400 --> 0:19:44.600 about to layer to those blocks. You can quantify degradation 0:19:44.720 --> 0:19:48.400 by blocks. You can match integrations, you can match reforestation. 0:19:48.880 --> 0:19:50.920 I asked Philip to imagine what it would have been 0:19:51.000 --> 0:19:53.880 like to attempt the tree planting project in an era 0:19:54.080 --> 0:20:00.399 before AI. His answer was, plant fifteen billion trees, restore 0:20:00.400 --> 0:20:05.160 the water towers. Impossible with Prithvy on Kenya's side, though 0:20:05.600 --> 0:20:08.960 it's really happening. What should be clear by now is 0:20:09.000 --> 0:20:11.600 how versatile Prithvie can be. I want to know how 0:20:11.640 --> 0:20:15.280 to combat deforestation. Prith vy can model that I want 0:20:15.280 --> 0:20:16.840 to know when the best time in the year to 0:20:16.840 --> 0:20:20.120 plant your crops is. Prithvy can help predict that too. 0:20:21.280 --> 0:20:25.600 Last year, six months after IBM started helping Kenya with reforestation, 0:20:26.119 --> 0:20:29.320 Kenya needed Prithvy's help on something else and it was 0:20:29.359 --> 0:20:30.160 an emergency. 0:20:31.000 --> 0:20:33.760 So something was happening in the world that we sort 0:20:33.760 --> 0:20:36.080 of had these flats that we didn't expect. 0:20:36.880 --> 0:20:39.520 In the spring of twenty twenty four, Kenya was hit 0:20:39.560 --> 0:20:43.840 with thunderstorms and torrential rain, days and days of it. 0:20:44.640 --> 0:20:46.879 And so I got a call from the Red Cross 0:20:47.040 --> 0:20:50.439 then one of my friends, and they're like, Ambassador, we 0:20:50.520 --> 0:20:53.639 need a little bit of help on how we deal 0:20:53.720 --> 0:20:56.400 with response because what we are seeing is unusual, right 0:20:56.440 --> 0:20:59.080 because no man, you would only have one area. All 0:20:59.119 --> 0:21:02.680 of a sudden, we had an entire country flooding. In April, 0:21:02.720 --> 0:21:07.280 we had about three hundred kilometers square kind of total 0:21:07.440 --> 0:21:11.760 land flooded, which is unusual for Kenyon. And so when 0:21:11.760 --> 0:21:14.080 I got this call, we were like, Okay, there's someone 0:21:14.080 --> 0:21:16.760 could did with IBM. We only did one function for 0:21:16.800 --> 0:21:19.600 the trees. It was actually a climate model, and we said, 0:21:19.680 --> 0:21:23.720 can we use this to help us better respond to 0:21:24.400 --> 0:21:28.000 floods and So that was how we started having this 0:21:28.440 --> 0:21:32.520 discussion with IBM in terms of repurposing the model to 0:21:32.640 --> 0:21:36.760 help us deal with this new challenge around floods. 0:21:37.320 --> 0:21:43.320 Again, prithvy is versatile. Prithvie could use everything it knew 0:21:43.320 --> 0:21:47.680 about the land, the forests, and infrastructure to analyze how 0:21:47.800 --> 0:21:52.000 and where and when floods would occur. The Kenyan government 0:21:52.040 --> 0:21:54.439 could then use the model to help the Red Cross 0:21:54.560 --> 0:21:58.920 organize its response, show areas that needed to be evacuated 0:21:59.240 --> 0:22:01.640 or safe place is with the Red Cross could set 0:22:01.720 --> 0:22:05.560 up camps. That information was invaluable. 0:22:06.520 --> 0:22:09.080 Historically, what has happened is that they would set up 0:22:09.160 --> 0:22:14.359 camp based on population congregation right where people assembly is 0:22:14.359 --> 0:22:16.800 where they set up a camp, not based on any data, 0:22:16.880 --> 0:22:20.440 right simply because people are there, they will come there 0:22:20.440 --> 0:22:24.000 to provide services and emergency response. What we realize is 0:22:24.040 --> 0:22:26.560 that that model doesn't work. So what we've been able 0:22:26.560 --> 0:22:29.000 to do with IBM is be able to to sort 0:22:29.000 --> 0:22:32.560 of give Red Cause very specific locations or options where 0:22:32.600 --> 0:22:34.960 to set up camps. So if people come here, just 0:22:34.960 --> 0:22:38.600 tell them no, move here, that's the safe place you 0:22:38.680 --> 0:22:40.320 really want to go. So I think for me that 0:22:40.440 --> 0:22:42.680 was really amazing. So we're calling them a very funny 0:22:42.680 --> 0:22:44.920 word for it, flood assembly points. We always have fire 0:22:45.400 --> 0:22:47.359 fire assembly points, but now we can say we have 0:22:48.000 --> 0:22:52.720 literally flat assembly points that are safe or citizens. 0:22:53.119 --> 0:22:57.679 That's fascinating. So the model has ingested this incredibly granular 0:22:58.520 --> 0:23:05.119 picture later of of the topography and weather patterns of Kenya. 0:23:05.240 --> 0:23:08.840 It's just giving you a set of useful predictions about 0:23:08.840 --> 0:23:10.800 how you should shape your response. 0:23:11.760 --> 0:23:14.240 Yes, and what we did remember is that, as I said, 0:23:14.280 --> 0:23:18.200 it was a full multistate called capability. What IBM gave 0:23:18.280 --> 0:23:21.000 us was a base map. We didn't have that before, 0:23:21.359 --> 0:23:23.920 and a base model. So you cannot have these layers 0:23:23.960 --> 0:23:25.359 up on layers, up on layers to be able to 0:23:25.359 --> 0:23:29.719 make intelligent decisions. 0:23:31.160 --> 0:23:34.120 Throughout my reporting on this episode, I've been really impressed 0:23:34.119 --> 0:23:37.160 by what Prithvie can do. But it doesn't stop at 0:23:37.200 --> 0:23:41.000 floods and reforestation. Prithvie has also been used to look 0:23:41.000 --> 0:23:45.200 at wildfires and floods in the UK, and Kevin told 0:23:45.240 --> 0:23:48.480 me that researchers in Africa have even used prithvy to 0:23:48.680 --> 0:23:52.600 identify locust breeding grounds, which could help them prevent swarms 0:23:52.600 --> 0:23:57.919