Saving the world one algorithm at a time | The Age of A.I.

Kurt Vonnegut said, "Science is magic that works." I buy. It makes a lot of sense... but it wasn't long ago that we couldn't understand what caused entire species to go extinct, why the earth shook or crops dried up. One of the promises of A.I. is that it will allow us to use machine learning for prediction and conservation, from protecting wildlife to anticipating earthquakes. Seeing the future may not prevent disasters, but I think we can all agree that we need an equipment upgrade. What the hell. Let's try it. Is a sixth mass extinction on the horizon? "Confusing answer, try again." Are you ready?

We're looking at the one with the necklace. There are around 250,000 parks in the

world

... ...and around 80,000 of them are threatened. This is a trail used by both elephants and poachers. We think so, yes. We have come here, all the way to Africa, to stop poachers before they kill using artificial intelligence. Fingers crossed this works. Poachers kill approximately 35,000 African elephants each year. A pound of ivory tusk can sell for $1,500 and the tusks can weigh 250 pounds each. Elephants are a decade away from extinction, but it is not just about protecting one type of animal or key species. There are around a million more species that are in danger of extinction.

Animals affect vegetation, biodiversity shapes ecosystems, all of which, in turn, impacts people. It's all connected. Therefore, it is no exaggeration to say that protecting African elephants protects humanity and our future. -A degree? - A degree. -Ten? - Ten? The Mara Elephant Project is a protection organization that includes active anti-poaching work on the ground and elephant research and monitoring. Here, on the grassy plains of southwestern Kenya, a team of just 50 rangers patrols an area of ​​more than 3,000 square miles. To keep an eye on the larger bulls... ...aerial teams tranquilize them, attach GPS collars and track them via satellite. - It also allows us to look at that historical data and say: these are brokers.

These are conflict zones. These are areas where elephants are at risk. Hoping to identify poachers, rangers have set up camera traps throughout the park. A traditional camera trap system takes a photo of anything that moves... ...which then results in all kinds of false positives. The hours of work of having to look through thousands of photos... maybe one caught a poacher, and it was three days ago. It's already too late for those animals. They are already dead. It is clear that traditional camera systems are not enough. So these are all different... Pitfalls. You have everything from barbed wire to cables, and then when an animal passes by, the trap closes on it. -These are for killing, for bush meat... -Okay. ...but you often find that elephants pass by too and you see that their trunks have been cut off, um, which is really sad.

If we can catch those poachers before they've killed the wildlife, red-handed with their traps as they approach, we'll really be ahead of the game. Each of these, if we had not eliminated them, costs the animal its life. My God. I'm the head of A.I. for Social Good at Intel. It is the conjunction of AI technology with social impact projects. They are throwing spears, specific for elephants. Oh my God. So what they do is they put a hoist on it and they hoist it up to the top of a tree, very, very high, and then there's a trip wire, and then when the elephant walks underneath...

It's disgusting. stuff. It is a problem that is not going to go away. What I do is talk to people and organizations that have a mission and connect them with the technology they need to carry out their mission. How are you? Hi Eric. Well well. Intel partnered with Resolve, an NGO. that uses technical innovation to solve some of the planet's most pressing environmental problems. Resolve is a small non-profit organization. We seek to try to save the

world

's large mammals and prevent the sixth mass extinction from getting worse. They began developing an AI-powered anti-poaching device they call "TrailGuard." TrailGuard at its core is a motion capture camera designed to help prevent poaching.

Thanks to the surprising invention of a new chip, it takes artificial intelligence to the limit, to the chip itself. So what we've done is put in this very powerful computer chip called the Vision Processing Unit. All the pictures that the camera trap takes will go through this VPU chip and it will determine if there is a person present or not... Okay. And I only send you the photos where there is a person. So it will reduce like 95% of the noise you're getting. -Wow. It has an A.I. Algorithm that looks at each image and sees if it's an image that the rangers are interested in, and at this point, they really just want to know if there's a person walking by.

The

algorithm

is fed with thousands of images of both humans and animals. It analyzes body shapes, facial geometry, movement and other characteristics until it learns to distinguish one from another, from any angle and in any light. It takes a photo every

time

movement occurs in front of it, but the A.I. It will send only images of people passing by. Image recognition involves looking at an image and understanding what it contains. It is the perfect example of something we take for granted. Humans can just look at an image and understand what's there without even thinking about it, but it's actually an incredibly difficult problem.

Surprisingly, we now have computers that, at least within certain parameters, can do this as well as humans. The beauty of this system is, here at Mara, the amount of

time

that passes from when the poacher walks in front of the camera until he says, "That's a human" and sends him to headquarters. Less than two minutes. We could then deploy a team of rangers to that site and hopefully make an arrest before they've had a chance to go in and kill those, those animals. We need to reduce the probability of detection... Correct. ...so that we don't miss anything.

We don't care if we get false positives, but we don't want to get false negatives, meaning a human passes by and the A.I. He didn't detect it. What I think worries me is that at night it will be even harder for you to detect it. One of the hardest things about implementing technology is that you are never exactly sure how it will work in the field. There are a million different things that could go wrong. Anna, Eric and the rangers head to the Mara Reserve to set up the cameras for their first real-world test. The parks in Africa are really big because the big mammals that live there need really big spaces to survive.

That really compounds the problem of trying to protect them... but the reality is that poaching raids are not random. In fact, poachers try to follow the trails because they want to get in and out as quickly as they can. This is beautiful up here. There are around ten main access routes that account for 80% of all poaching traffic. I think this is where the elephants usually come. What could go more unnoticed? -What do you think? - Mmm... I want to hide it behind this branch... Yes. Excellent. This looks amazing. So, you have the view. Poachers wouldn't even suspect that there is a ranger or a camera here.

This is a hidden ranger. - A hidden ranger. - Yes. So tonight, my friends and I are going to play cat and mouse with the cameras they brought out. Let's imitate real poachers. Let's see if the cameras catch us. Does artificial intelligence work at night? Which is essential, because a lot of poaching occurs at night. Night is also more difficult for image recognition. Details and features are harder to see, so for this test, the A.I. You will have to rely on a limited data set. We knew that it works in the laboratory, in the United States, but it is not the same.

This is a trail used by both elephants and poachers. We think so, yes. 19 kilometers away, at MEP headquarters, the team waits for a response from TrailGuard. Very good, well, here I see that four images have arrived. AI computer vision has detected something. Person, animal or other? So, let's review the first one. Fingers crossed this works. - Hey! - Yeah! There is. Well done guys, well done. - Awesome. - We have our poacher. I'm like on the moon. I'm very excited that the demo worked. - Oh great! - - - Two by two. Check it out! Yeah...

Oh, it's so good. Well, four by four. That's pretty good. Catching poachers has actually been a bit of a cat and mouse behavior, so to speak. TrailGuard is really giving us this new advantage. We will catch the poachers now, before they have a chance to come in and kill the wildlife. This is a global problem. There are far more bad guys than good guys defending these parks. Right now, we can help tip that balance toward the good to save the world's large mammals and prevent the sixth mass extinction. Mass extinction occurs when most of Earth's animals and plants die, wiping out biodiversity.

Probably the best known was 66 million years ago, when a giant asteroid crashed into Earth and decimated everything. We can now say that we are in the middle of a sixth mass extinction. Poachers are part of the problem, but so is climate change and what we eat. - All of us. Can we reconfigure our destructive behavior? Sergio Barroso is one of the most talented chefs in the world. He has taste buds that no one else has. Well, how about this one? What do a Chilean chef, his incredibly sensitive taste buds, and a robot named Giuseppe have to do with

saving

the environment?

Good question. This food could be the key to

saving

our planet. It is the definitive test. We have very ambitious dreams. Yes we want to change the world, but we are not dreamers... We are rather doers. So let's add 90 grams. Animal protein foods have been our main source of nutrition for the last thousand years. I actually grew up eating steak, but the way we grow it...isn't the right way. The food industry has become the common denominator for all the major environmental ills known to humanity... deforestation, water scarcity, world hunger! Incredible true? Does eating meat cause climate change?

Sure. Cows and other animals emit methane, a harmful greenhouse gas. One third of the world's arable land is used to feed livestock. Look at it this way: Eating a hamburger has about the same environmental impact as driving a gas car for ten miles. So what do we do? There is a new way, a better way to prepare food, and it is plant-based. Matías teamed up with two PhD friends to co-found NotCo, an artificial intelligence company. Start-up with a humble mission... Save the planet by reducing meat consumption. So we can see here that most of the ingredients make a lot of sense.

Cabbages, rice, pumpkin, sunflower oil, pea protein. His challenge is not to create a plant-based alternative to popular animal proteins. Many companies already do it. So there are a lot of formulas being generated here and they seem to make sense, right? NotCo is trying to figure out something a little more elusive... It looks like it's using three different oils. ...taste and perception. Using an A.I. An

algorithm

they call "Giuseppe" tries to make people think they're eating meat, eggs or milk, when in reality... they're not. I was trying to find omega-3 from a mixture of three different oils. ...a technology that is capable of telling us how to reproduce a food of animal origin simply using plants.

The algorithm is able to understand that there are clear connections between the molecular components of food and human perception of taste and texture. The magic behind Giuseppe has to do with chemistry. AI analyzes the molecular composition of foods, such as milk. Then create a list of ingredients from your most basic building blocks. Finally, using machine learning and a massive database, Giuseppe recombines select plant-based food elements to recreate the taste and texture of the original. Humans are good at reasoning about two ingredients or maybe three ingredients at a time, but after that, it becomes very difficult for us to think about it, but the machine can start thinking about five ingredients, ten ingredients and how they all go together . and what the flavor profiles will be, and that's really the great power of the machine.

What are you doing? Are you working on fibers? There are many similarities between plants and animals, because we share part of the chemical nature. Everyone has DNA, everyone hasRNA, they all have proteins, lipids and carbohydrates. An almond and a walnut share, like, 97% of the molecules are the same. Only 3% gives your brain the identity that a nut is a nut and an almond is an almond, so we need to really identify the molecular characteristics that give food identity to identify the specific type of plant and the specific type. plant ingredient to rebuild. Yes. It smells like sushi.

Not fresh sushi. I mean, that's the magic behind NotCo. We're building an ecosystem that will really look at food in a unique way... because Giuseppe suggests really... crazy ingredients sometimes. The first food they tried to recreate is one of Chile's favorite condiments... mayonnaise. We identified that cabbage, in a specific environment, will release a molecule that is very similar to lactose, and for the brain it is more or less the same. We tried the emulsion and said, "It tastes exactly like mayonnaise..." but it's red. The algorithm did not yet understand that one of the characteristics we value in the sensory experience of a product is color.

So they solved the flavor and texture, but kept tweaking their algorithm until Giuseppe came up with a better formula for the color. We are the third largest supplier of mayonnaise in Chile and the best-selling mayonnaise that comes in squeeze bottles. The success of Not Mayo led to the creation of Not Milk, Not Ice Cream and Not Meat. Now they are working on Not Tuna. We have devastated oceans. Generating a substitute for tuna is something that will move the needle in the world, so fish makes sense. Victor, if you can go to the berries and extract the color, please.

We need to make a bluefin tuna. It's the... normal tuna. We have a cabbage that is prepared like kimchi. It is the appearance, flavor, aftertaste and other dimensions of the sensory experience. That's the key, man. We can add this to make an emulsion. If we don't do things that are as good or better, things won't change and we won't move the needle. So this looks like tuna color. -Yes, very good color. -He's very focused... Yes, I think that's good. It's quite similar. One of the things that comes with large-scale behavior change is that it's often not driven by nutrition or sustainability considerations, but rather it's driven by taste.

I think everyone has the experience of wanting to try the new taste of potato chips, and if we can have that same ownership in healthy, sustainable foods, that can be really powerful. - Hey. - Hello. These are the first formulas of Giuseppe Tuna. Today, the NotCo team is testing two of Giuseppe's tuna recipes. What do you think? It smells quite fishy. You know, it's like, it's like, um... It's like this smell of ceviche, like, and some seaweed... It's really the first time we've tried fish at Giuseppe. The color comes from the berries and kimchi we use.

Wow! The texture is good. MMM. It lacks a little flavor. The algorithm is not yet ready to understand the complexity of a muscle. How do we explain that to an algorithm? Can we try that one? Hmm. I like this one better. The taste of this is better. -The flavor is much better. -Flavor... But the texture of the tuna is this, but that one has a better flavor. Wow, I'm impressed for being the first shot. When Giuseppe suggests something, it's the first iteration, right? We need to keep executing this. I mean, that's where scientists take control of the process and try to really delve into the molecular and structural component of food.

In terms of cans, I think we're close, you know, we're very close. Not Tuna is the new Not Chicken of Not Sea. 92% of our consumers are not vegetarians. They don't care about sustainability. What matters to them is eating delicious food. Everything Giuseppe suggests, all the recipes, all the products end up here, and this is basically our R&D and product development facility. Sergio Barroso is one of the most talented chefs in the world. That's not a euphemism. Barroso got his start at El Bulli in Spain, arguably the best restaurant in the world, and his own restaurant in Chile is in the top 50.

He is a culinary innovator, can taste flavors invisible to the average human language and carries with him his own personal style. spatula, gunfighter style. Oh, you have your spatula. Yes, always, always. You never know when you have to try something. Suffice to say, NotCo couldn't have chosen a more discerning palate to put its latest A.I. foods to the ultimate test. The most important part for me is the flavor, because without flavor we don't care about the other part. The acidity and creaminess are very, very balanced. You would never think this is doing it without eggs, you know?

Well, now Sergio, you will try No Milk. Well. Health. Health. First, the texture is like milk. Yes, it is very tasty. It's very, very, very tasty, and I think it has more flavor than the milk you can buy at the supermarket. So Not Milk is more milk... -Yes. --than milk. Milkier! We started with a plant-based burger, because we think a burger is not as complex as a steak. The texture is the same, just like when it is cooked. When I see something like this, part of my mind thinks "What...what can I do with this", right?

So I want to try, cook with these, with milk, with meat, with mayonnaise, with everything, right? I think I can prepare something very, very special. Fantastic. Challenge accepted. Health! -Holy! -Sante... This was planned a day before, so Sergio is with his team, preparing an eight-course dinner for the very select group of people. Expectations are very, very high. Anxious, nervous, you know, having mixed feelings of excitement. We are going to eat everything the same way we prepare it in a normal recipe at the restaurant. We would put a liter of milk, we are going to put a liter of Not Milk.

This is the last challenge for us. At the end of the day, if NotCo is going to take off, their food has to taste good to everyone... A.I. or not A.I. I like to start the menu like this... We want to say in the first cover that this is something serious. Oh Lord. Oh...I couldn't believe it wasn't mayonnaise. He is using Not Meat to make dumplings. Normally with real meat you need three minutes, but you don't know it, because it's the first time you're cooking with a meatless meat. We are very nervous because this is going to show that plant-based foods can actually replace animal-based foods.

Ladies... My brain told me I wasn't eating meat, but my soul told me it was meat. We are going to prepare the last dessert with No Ice Cream. We are making something similar to a sweet taco with the cotton candy. They weren't thinking all the time that they were eating something different with "No" products, and they were happy with the flavors, with everything, right? - - It makes me feel fantastic. Saint. Thank you all. Thank you. -Very good. Very good. -Thank you. We want to change the world. The dream is big, the dream is there. NotCo's dream is for it to use A.I. to preserve our planet from the damage humans are causing... but is it effective both ways?

Can AI? Protect people from the destructive forces of nature? The laws of nature do not change. Humans change in response to AI... but nature does not. If we take earthquakes, if we can predict when they will occur... this can save many lives. It has been 319 years since the last major disaster here. Seismically, it's the quietest area in the world right now, so the possibility of it being quiet because it's locked, loaded and ready to go, has certainly spooked a lot of people. 25-238, we have an accident with serious injuries. They are advising serious injuries. We have a 50-foot section that has collapsed.

If a magnitude 9 earthquake occurred on the Cascadia Fault, we would expect tens of thousands of casualties. Right now, we can predict that large subduction zones or major faults will experience an earthquake at some point in the future, but that's not what people want to know. Do you want to know if the earthquake will be imminent? In six hours? In a day? In 30 seconds? The Cascadia Fault, more than 600 miles long from Vancouver Island to Northern California, could cause the largest natural disaster in North American history. It is generally accepted that The Big One is coming, perhaps within 50 years, but experts can't be sure or be more specific.

The PNSN is a network of seismic sensors, and the idea is that we are continuously monitoring seismic activity of all scales, at all times. This is one of 400 seismic stations in Washington and Oregon. This is a strong motion accelerometer, so it looks for large earthquakes. The earthquake early warning system does not predict an earthquake. We're doing a stomp test. I stomp on the ground, the instrument detects that tremor. It's about identifying those first waves that are arriving. Our 400 instruments send a constant stream of data to our data center. In each of those 400 sensors there are many signals that are not earthquakes.

So we're generating this incredible volume of data. It's probably a passing truck. Where the A.I. enters would be to filter out what we call cultural noise. Trains, trucks, people. Information about vibrations from the region's 400 sensors is fed into a machine learning algorithm, which is trained to differentiate earthquake tremors from, say, construction or bus tremors. Using machine learning and a huge database of known sounds, the A.I. You can quickly classify noise from the natural world. Oh, there are some interesting things happening here, guys. Machine learning allows us to quickly find the signal of earthquakes. We can do it much faster and better.

We are working hard on an earthquake early warning system and it is called Shake Alert. The emergency management center here would be one of our key immediate users. - -Earthquake! Earthquake! If it's a big earthquake, within a second or two... Unit clears in five, identify yourself. ...our computer algorithms determine which area is going to be affected... Damage Assessment... ...and then create a warning for that. Depending on how close you are to the source, the warnings can be very brief, from less than a second... ...to perhaps up to three minutes. ...but when a massive earthquake struck northern Japan in 2011...

Residents received a warning... between 20 and 90 seconds, depending on how far they were from the epicenter of the earthquake... ...and still Thus, almost 16,000 lives were still lost. Is there a way for A.I. Could it give us even more warning time by predicting the next big disaster? Earthquake prediction is a really difficult problem, because we don't know where earthquakes will occur. Not really, right? We don't know when they're going to happen, and we don't really know how big of a zone the fault reaches before it suddenly ruptures. Well, what do you think, friend? Yes, we can go ahead and make a compensation right now.

Yes, let's do it. Chris and his team have come up with a way to create earthquakes in the lab. You're at 2.3 right now. We're simulating the kind of things that happen in the upper few kilometers of the Earth's crust. So I guess I'll set this up before we start. The goal today is to try to find out if we can create laboratory earthquakes in a range of conditions and use machine learning to predict that entire range. On a small scale, this test mimics what happens on a massive fault like Cascadia. These are basically two massive tectonic plates, one under the northwest coast and one under the Pacific.

One slides under the other and sometimes catches, causing friction and pressure. When it gets too much, they slide violently, causing an earthquake. So that's one side, and then I basically repeat the same thing for the other side block... That's what Chris and his team are trying to replicate with mini granite blocks in the lab. We have laboratory seismometers located right next to the fault zone. ...and I try to align it with the lasers. Giant, home-built, hydraulically powered press. Now this is ready to go. I'm turning it on right now. In the lab, we control the stresses at the fault, we control how quickly the stresses build up to some critical point where it will rupture.

There are moans and creaks from the fault. Microearthquakes, many small mini-fault events. Yes, let's see. Let's look at the acoustic side of this. We are listening to everything that happens in the failures and using machine learning to find the patterns in it. Is there something happening that we can use to say, "Oh, this is about to be big." -Yeah. -There is another? So now we're recording. We're listening to all that small talk you hear. Let's zoom back in, because we're going to have a big one here right now. Here it comes. -Pop. -Here we go. Yes.

That was great. That yellow line is the measure of the shear stress at the fault. We're seeing it grow, and as it grows, there are microearthquakes that we're hearing, and what we've realized is that we can use those microearthquakes along with machine learning to predict the timing of the next one. event. These microearthquakes you're hearing are too weak for human hearing, but not for machine learning. Uses weak signals to predict earthquakes thatwe can hear and feel. That was a nice big high frequency event. We'll probably get another one... -Ooh, that one was big. -That was great.

Listening to the small events to teach us when the big events will occur. No, that's fabulous. That's so cool to see. Another one approaches. We are predicting the time of the next earthquake and its duration. How many cycles do we already have? Something like 20, 30 events. It's a game-changer to use artificial intelligence, because now we can use machine learning to ask questions about why... why is that happening? Is there some geometric structure being built within the fault zone that is somehow seen by artificial intelligence? Machine learning prediction is here. The red line is the experimental data. Every time it falls, it's a laboratory earthquake, and then the blue line is a model prediction based on machine learning.

It's impressive how well it works, right? Yes, you see the same thing in different cycles, exactly the same group appears just before the failure. Yeah, you can see, you know, it's not perfect, but it's very close to perfect. We have proven beyond a doubt that AI, machine learning, can predict earthquakes in the laboratory. We'll probably get another one. Yes, there it is. That's great. Yes of course? The challenge is how do we scale that from what is measured in the laboratory to what we measure with seismic sensors over an area of ​​hundreds or thousands of square kilometers.

As artificial intelligence moves from the lab to the world, the world becomes much more complicated and therefore there are many factors that may not have been modeled. If we can transfer, if we can generalize, it would be really amazing. How many years will it be before we can predict or forecast anything about Cascadia? I think that's within our lifetime. It's a bold statement, but one that no longer seems out of reach, especially as machine learning gets better at identifying patterns to better forecast earthquakes or other natural disasters... and if A.I. can predict a calamity, could you go a step further and prevent it?

Corporations and governments are using machine learning to help resolve large-scale conflicts and catastrophes. One area that causes many problems is diet or lack thereof. Agriculture is just one critical component of national security and the health of the world. Food shortages sometimes cause famines. Hunger leads to political unrest. The way I see A.I. It is not a terrifying force to fear, but something that will help us address these big problems. Mark is the co-founder of a company called Descartes Labs. This is a maximum combination of what the detector detected last winter. This year the whole place lights up.

We are a young company. It's me and my merry band of physicists. One of the most interesting things we do at this company is trying to look at the latest developments in science and trying to figure out how it can affect our lives. There is a huge amount of satellite images out there. That's one of the largest data sets humanity has ever collected. The most surprising thing to me about satellites is that we have been receiving excellent scientific information since the 1970s, and yet it is really difficult to use that information. There are thousands of satellites photographing the Earth, so Descartes built a supercomputer in the cloud that uses machine learning to analyze these images and create models from the information.

They are trying to predict when diseases, disasters, or even wars might occur. Artificial intelligence has come a long way in the last decade and now you can start building models from this data instead of just having a bunch of unwieldy images. We have been trying to find all the solar panels in the US using these satellite images. So in the first task, we build what we call a similarity search engine. The search engine uses object recognition, a type of A.I. that learns to identify and differentiate specific things within images. Instead of looking for poachers in Africa, they look for everything from solar panels to river beds.

The computer only sees numbers. He doesn't see the image the way we see it. He just sees raw numbers, so over time we teach him what a river looks like. That's a street. That's a building, and the algorithm is able, over time, if it sees enough examples of this, you know, it says, "Yeah, I get it." So we started looking for a problem that we could solve with this, and what we decided on was agriculture. Plants are really pretty because they're like these little factories, right? Just by observing how the light bounces off these plants, you can tell a lot about the production of this factory.

For innovators, it's not totally unusual to find an answer and then reverse engineer a question. The question Mark asked his A.I. It was corn. How many corn fields are there, where and what are their growth patterns... and then by comparing satellite images over time, can we predict how much corn the country will produce next year? At the beginning of the project, seeing if it was possible was something of a pipe dream. Could you use sensors that fly hundreds of miles above those corn fields, without ever having seen an ear of corn, and get a really good, accurate prediction?

We combine two things, so the satellite data will give you an idea of ​​the health of the plants today. We then analyze weather data, which will give you an idea of ​​how healthy the plants will be in a week or two. In 2017, the United States produced more than two billion bushels of corn. Descartes' estimate was within one percent. ...and I think what really surprised the industry was that it was not a group of agronomists, they were not people who were experts in corn. We were a group of physicists who simply used the principles of physics and light.

This really woke up the market that data can really change traditional forecasting methods. If you have a lot of data and you have an algorithm that can analyze it, it can learn things that no human being could perceive and it will be able to make predictions that no human being would ever have made. able to make. ...but it's not just about corn. You can do this with almost anything photographed from space. Water, forests, factories, roads. I'm thinking about mapping rice paddies in Asia. Brighter yellow has a higher chance of being rice, and any darker purple or blue has a low chance of being rice.

Can you know the health of a crop? One of the ways to understand the health of crops is to look at the infrared bands. This is beyond red on the electromagnetic spectrum, which says a lot about the health of crops, but humans can't even do that. This is in Iraq, where they grow rice. Most of the water in this area comes from melting ice. There are many places where the food supply depends largely on the amount of snow that falls during the winter. The blue here is like looking at the snow from 2014, and 2015 was a drought year, and there was much less snow, and it correlates with how much rice there was in this region. 64% less rice.

So if we know how much snow falls in winter, farmers will be able to plan their crops better. Maybe they plant drought-resistant crops. This should give you an idea of ​​how much rice that area can handle later in the summer when the snow is melting. Its AI-based forecasts raised eyebrows and caught the attention of DARPA, the defense research arm of the US military. DARPA is the Defense Advanced Research Projects. They analyze the most difficult problems that will affect the United States military. We've been pretty focused on the science, but I know you work very closely with the defense department.

Do you think this is going to influence their thinking about conflicts that may arise, or... Oh, absolutely? I think having advance warning about problem areas is... it's very critical. You know, the Arab Spring was caused by a wheat shortage that caused a bread shortage. No food, no nothing! Hosni Mubarak. The problem we are faced with is looking at food production in the Middle East and North Africa, and the goal is to understand where there might be food shortages. When you don't have water, when you don't have food, when you don't have shelter, you fight, right? You will do anything to survive.

I really hope to receive an update on the current status. Well, we have been analyzing the current year's growing season. This year we are seeing many healthy wheat fields in Syria. It looks like we are about 20% above last year's production in Syria. -This is a wonderful advance in the ability to obtain, you know, objective measures of what is really happening in agriculture in these problem areas. It is very important to know where crops are failing so that aid can be sent there before the situation worsens and turns into famine. The fact that we can look around the world and find where famine could occur four months from now is... it's mind-blowing.

I am quite surprised that the health of crops can be observed with satellites that fly hundreds of kilometers above the Earth. The idea that corporations and governments have technology that predicts, and perhaps prevents, large-scale human catastrophes, such as war and famine, is mind-boggling. Computers are like three-year-olds right now, and we are training them to be our aids, to help us make better decisions, to help us be better humans. I deeply believe that science will help us save the planet and save ourselves. In the old days, people used to think that disasters were caused by God or magic.

Now we know better. If a sixth mass extinction occurs, it will probably fall on us. We have this technology that we can use to save life on Earth. AI. We may not prevent disasters, but new scientific tools such as machine learning, image recognition and predictive models could at least help us get ahead of them. Artificial intelligence and machine learning will be a very, very powerful tool to make predictions with precision that was impossible before. There have been many sustainability projects that are implementing artificial intelligence to address the problems facing the world. If we can get a better idea of ​​what the future of food will look like, then we can really save a lot of people.

Conserving the planet or preserving our species really has nothing to do with magic... but it would be divine.