Science and innovation for sustainability transformation strategies (1/6)

MARIETTE DiCHRISTINA: Our first speaker I’d like to welcome the stage is Dr. Bernard Meyerson. And he is the chief innovation officer of IBM. Dr. Meyerson will give a keynote on how science and innovation can lead to transformative business strategies. After he shares his prepared remarks, we’re going to invite you to ask your questions, and we can begin what we hope will be a lively conversation for the day. Please join me, Dr. Meyerson. [APPLAUSE] BERNARD MEYERSON: I want to thank the organizers for inviting me. And it’s incredibly fitting to do this in Singapore. I mean, Singapore is the model of sustainability. I’ve been working in this nation on and off for 25 years, with the government, with various private agencies. And it’s just astonished me that you take a country that has virtually no natural resources– I mean, you haven’t got enough native water, much less oil or anything else. It’s only got really, one natural resource, which is its own people. Well, it turns out that’s been enough to compensate for the lack of everything else. And in fact, it’s an incredible success story. So it’s really very fitting that we do this here. Now I’m a hardcore scientist, so believe me when I tell you, arguing that science and technology are a necessity to get this done comes very naturally to me. And in fact, I had the incredible privilege and thrill and occasionally torment of working with my dear colleague, Mariette, as part of the World Economic Forum, where we basically identified the top 10 emerging technologies as part of a forum effort for 2015-2016. These were technologies that we felt would become essentially groundbreaking globally, and they would emerge in common usage within the three to five-year time frame. The reason I pulled these up is because as I’m arguing from the beginning, in the broadest sense of sustainability, it is driven by science, technology, and innovation. To make the point, what I did was I basically– and just there’s the list of the top 10 for each year. Those in the back, if you have binoculars, it’ll work great. But what I really wanted to make the point of is not the list so much as if you look at that list of things– and I’ll pull off a couple just for those who can’t see them– fuel cell vehicles, right? Additive manufacturing, sense and avoid drones, neuromorphic computing, precision genetic engineering, CRISPR-cas, nanosensors, next generation batteries, blockchain, organs on a chip, perovskite solar cells. I mean, these are basic scientific and technical breakthroughs that were done. The interesting thing is if you look at them in the context of sustainability– because remember, the thesis here is that sustainability is derived through science and technological breakthroughs– they are the ones that make, really, a difference when it comes to what you can do in sustainability. If you just highlight in green the ones I think are directly applicable to topics relevant to sustainability– the vast preponderance. The vast preponderance. And if you start to break it down a little more granularly by topic– this is my favorite one, global warming. OK, how many here actually believe global warming is a real phenomenon? Thank you. Good lord, I was afraid I’d have to go out there and strangle someone. [LAUGHTER] You know? You’ll forgive me. This is a comment I will make on behalf of myself, not my company at IBM. You have got to be kidding when you’ve got a guy in the EPA who says the Earth isn’t warming up, but I am buying property 10 miles inland nowadays. Yeah, you betcha. This is a real challenge. But when you look at global warming, out of the things up there, at least four of them are directly related to dealing with it. Fuel cell vehicles basically use hydrogen versus hydrocarbons, and potable water is the output of the exhaust pipe. I’m not suggesting you hang upside down from the rear bumper of the car and drink from the tailpipe. But at least what’s coming out is not CO2. Now there is a catch. Be real careful. I’ve seen people say this is great. But the way they’re generating the hydrogen produces so much CO2 that you’re better off just burning gasoline. So be a little careful with some of these things. Neuromorphic computing is a completely different way of organizing a computer, where you dramatically reduce the kind of power required to do basic things, for instance, look at a video and identify things of interest. We’re talking factors of 1,000X. As you’ll see soon, it matters. To give you some idea, in China for security reasons, the state has deployed 200 million video cameras. Believe me when I tell you bringing all that data into a central point to analyze the image is a waste of energy. Why? Because you can’t. There’s just too much data flowing. You need intelligence out of the camera so you only bring to the center what’s of interest. Looking at a parking lot where nobody is moving and sending the video to the center for analysis is a waste. Things like neuromorphic computing take a factor of 1,000 out of that energy usage. Major effort. Next generation batteries– well, electric vehicles, obvious impact. Perovskite solar cells, very, very inexpensive, a cheap way to make solar cells. Again. Other topics, aging in place. This is a major issue in societies in Singapore, Japan, where you have aging populations– in the US, the baby boomers. The technologies that were called out here address that. Autonomous vehicles for those who can’t drive. Nanosensors for local health care, and so on and so forth. Again, these are major factors, major players in those areas. Inclusion I consider an issue of sustainability. If you are not inclusive in a society, if you’re not inclusive globally, what will happen is you leave people behind, you have an unsustainable situation. You cannot have this huge divergence between the haves and the have nots. That is not acceptable ethically and is not sustainable socially. When you look at things like additive manufacturing and distributed manufacturing– additive manufacturing is basically 3D printing, as an example. The point is you can have that virtually in any location and bid on manufacturing work all over the world over the web, and literally just send your result back in from a distance. The point is things like this actually are emergent. And because they are not capital-intensive– it’s not like building some huge factory– you can actually do this on a remarkably small scale. And there are astonishing examples already of people who have leveraged this to build things in locations you would have never thought to even look. Nutrition, health care, precision genetic engineering, CRISPR-cas, the ability to have a true gene editor, not a crap shoot. You’ll forgive me, but frankly– that by the way, is a reference to casinos, nothing else. A crap shoot basically is what we currently do for genetic modification. You basically take a virus, and you hope when you stick a piece of genetic material in it, that the virus inserts it somewheres where you want in the DNA of what you’re attacking. This is not a great thing. Because where it winds up can make the difference between you have three arms or a desired thing. Genetic engineering, where it’s truly engineering– CRISPR-cas is a real editor where you can actually snip out and/or replace a piece of genetic material at a precise location in the genome. This enables you to do things like take rice and potentially develop a type of rice that may require 30% less water for the same yield, or 30% more yield for the same water, which is the difference between living and dying, starving and having food. This is an important element, and again, sustainable. Infrastructure– some of the stuff here, if you look at it– thermoset plastics, for example. Basic science was done, because they weren’t recyclable. And they’re in virtually every automobile in quantity. If you look at the news yesterday, somebody was estimating within the less than five years there’ll be more plastic in the ocean than we have fish. OK, that’s terrifying. And if you don’t do something about it, it’s going to have dire consequences. The fact of the matter is that the thermoset plastics which were not recyclable– a woman took it upon herself– a woman chemist on the west coast of the US simply looked at it and said, it’s not acceptable. And she actually redid the underlying chemistry so you can break them down to the basic monomer. And you can now recycle them completely, just like that. Again, sustainability is driven directly off of the work of scientists and engineers. Similarly, two-dimensional materials– what does that have to do with infrastructure? It’s literally a one atom thick layer of material, generally carbon-based. But in the plane of the material, it has unimaginable strength, which means if you mix these in in the proper quantities, for instance, with cement, you can generate a very light but unimaginably strong material, thereby requiring far less cement to build the same structure. That has huge environmental implications. Because cement is energy intensive beyond all imagination to create. Again, basic science leading to sustainable outcomes. Now the interesting thing is a bunch of these are completely dependent, however, on one of the elements I already identified, which is AI. AI and machine learning are sort of the hot topics of the time. And of those things up there– it’s hard to see which ones I’ve identified– but as an example, autonomous vehicles are heavily reliant on this. Nanosensors and the internet of nanothings– if you’re going to use it broadly to look at health, perhaps look at the outcomes connecting the genetic code of somebody to the treatment they’re hit with, to the result of that treatment– all of these rely upon taking the vast amounts of data that are coming in off of these various systems and turning it into knowledge. And so I’m going to focus down on that one. But before I do, I just wanted to be complete and point out that Mariette is now chair and I’m vice chair, which causes her to take all sorts of advantage of me. But nonetheless, in 2017, this was taken over by Scientific American on an ongoing basis. And here if I just circle in green– if you look at the list of the top 10, in fact, at least six of those 10 are directly related to sustainability. In other words, these are science and technical achievements that are emergent that nonetheless have direct reading. And in fact, one of the six– it’s a bit ridiculous– is the sustainable design of communities. I mean, that’s not that subtle. But the point is I think I’ve beaten this point to death. I don’t think there’s really a whole lot of need, particularly because this audience is clearly fairly technical, to convince you that science and technology are the underpinning. So now I want to go a little deeper and A, get into some of the underlying science, and also some of the more– how should I put it– the more around the edge effects this has on sustainability, and some of the more, shall we say, exotic definitions of sustainability that we have to pay attention to. First and foremost, research created the foundation of this science and technology when I’m referring to AI and machine learning, and sort of the cognitive era. But it didn’t start with a poof. It started with a huge investment. Because we had to first handle big data and analytics before we could go on and actually start even talking about AI and machine learning. Defining big is a bit of a challenge. Now this is a personal medical question. Feel free not to answer it. I’m going to, because it’s not that magic. How many here have had an MRI? Yeah. Boy, you guys are the problem, like me. You have to understand the amount of data generated when you have an MRI would drown a typical system five years ago. We’re talking basically terabytes of data, at a minimum. And the thing is right now, the estimate is when we all basically, over the lifetime have all the tests, all the video and imaging and all that done, we will each generate the equivalent of about 300 million books of data, 300 million each. Now that’s a scary prospect. Because let me give you some perspective. The Library of Congress– pretty much among the largest libraries in the world– the Library of Congress in the US contains 186 million copies. So basically, you guys are about one and half versions of that each. This is a really big problem with data, particularly when you do stuff generated by the internet of things. And I’m sort of grouping that kind of data underneath that header. The scary part is currently, the data from the physical world, including us, is growing by about 44 exabytes– 10 to the 18th zeros after this– a month. That is a terrifying quantity of data. And the bottom line is– and it’s very, very interesting– this is basically a fundamental issue, because you can no longer store data. All of our careers, we’ve been taught, we’ll just store the data and we’ll analyze it. No, you’re not. Data is just not dense enough, and most of it is garbage. What we are going to have to start to do is think differently. We will have to start to store knowledge so we understand the data. Just storing the data is a waste of time if you’re not going to do anything with it. And the vast majority of data out there today you do nothing with. If you don’t understand data that you’re looking at, the price of that lack of understanding is dramatic and terrible. I don’t care whether it’s the physician on the left side who’s looking for a brain bleed due to an aneurysm that is less than 4 millimeters across, where the odds of doing it manually are about– this goes back a ways– about 10% of seeing that brain bleed due to that small an aneurysm. Because they have to look at an MRI with 100 slices, staring at these things without any guidance is misery. Whereas if you understand it, or you have help from a system that’s smart enough to at least circle suspicious regions, the odds of seeing that same phenomena go to 90%. Similarly, in the middle, if you don’t understand the vibrational and other data coming off of pumps deep under sea in an oil system, you can have a catastrophic environmental effect, or you could collapse the economy of a nation by fraud or stupidity. Sometimes it’s a little mix of both. The point is you really do have to understand data. And we learned this the hard way. The sustainability problems that we face are because the data comes at such huge volumes, there is a question as to its veracity. Because not all data is correct. And of course, the velocity at which it’s arriving is terrifying. You have to overcome all these things. And there’s a good example I’m about to show you as to why it’s so incredibly important to really understand the data, not just to collect it. [VIDEO PLAYBACK] [MUSIC PLAYING] – This is a baby, a baby generating data in a neonatal ward. Every heartbeat, every breath, every anomaly, from over 1,000 pieces of unique information per second, helping doctors find new ways to detect life-threatening infections up to 24 hours sooner. On a smarter planet, analyze the data and you can predict what will happen faster, so you can do what they’re doing in Toronto and build a smarter hospital. [END PLAYBACK] BERNARD MEYERSON: It turns out that in the neonatal intensive care unit for premature infants in the Sick Children’s Hospital in Toronto, they were collecting 2,000 data points a second from each child. And yet they constantly experienced something called baby crashing. It is where the child becomes septic but there are no warning signs. In all the data they were collecting, the heart rate, the blood gases, they were all in the normal range until they fell off a cliff and the child died in minutes. And this was maddening to the professionals involved, because they could not see in the data what was causing this. So they called upon us, and we worked jointly with them and the institute. And what we did was we collected the data for the child as long as that child was in intensive care. Now these are preemies. You understand the problem is they don’t have well-developed immune systems, so it’s not like you or I. If you get a cold, you start sneezing and hacking, blowing your nose. The point is, that’s not the disease, that your body’s reaction to the disease. That’s the warning sign. These kids– if you don’t have an immune system– nothing was warning you until literally, the sepsis overwhelmed their core organs and they died suddenly. We kept all this data. And god forbid, this child died. We went back through the data saying, what on earth did we miss? And it turns out with sophisticated analytic tools, we found something remarkable, which is, for instance, the heart rate was in the right range. There were no alarms saying the heart rate’s out of range. But the time rate of change of the heartbeat– in other words, the speed with which it was changing– because in preemies, the heart rate wanders all over the place. It turns out it wasn’t wandering as much. That was one of the high signs of impending sepsis. Similarly, if you matched that with a second order effect in changes in blood oxygenation, which were tiny and imperceptible to the human eye– when you took all those things together, we could predict sepsis 24 to 48 hours, as it turns out, before it hit. Which meant you could use antibiotics. You could use antivirals, which are heavy duty drugs you wouldn’t normally give a preemie. But you know what? It beats the hell out of having them die. OK, this was a real eye-opener for us, because this is seven, eight years ago. This isn’t current. But what it taught us is data is useless without understanding. Understanding it is real power. So it’s really about turning data into knowledge. That is the fundamental issue that was getting in the way here, is a very limited ability to do that. So this is one of those wonderful examples of just a learning experience, and why we focus so much effort on AI and machine learning and all the other attributes. Because doing this manually takes forever. It is an astounding outcome, but it isn’t something you can do quickly if you just do it by hand. A similar challenge– video at scale. In London, they have, they estimate, 500,000 video cameras. So they take about 300 images of you every day if you’re just wandering around London. I think I said in China, there are now 200 million of these. The point is, part of sustainability, of course, is being able to assure safety. The challenge, of course, is when you have video systems, you also want to ensure privacy. And it’s a very delicate balance. You have to be very careful here. Now this may involve dealing with data at volumes that are unacceptable, as you can imagine. In point of fact, that in London comes to about 10 to the 13 bits of data a second flowing to a central point. This dwarfs any ability to deal with it. So what actually happens is you’re not turning the data into knowledge. So these systems are typically, unfortunately, used after some disaster has occurred to go back and see why, as opposed to predict a disaster will occur and prevent it, which is really what you want to do. Now at the beginning, I talked about there being neuromorphic computing takes a factor of 1,000 out of, in fact, the energy required to analyze a video signal. Why is that important? Because what you’d really like to do is put one of these little neuromorphic systems on every camera, which burns very, very little power, and only send pertinent data to the center for analysis. You don’t need a picture of a parking lot when nobody’s moving. And you don’t need a high-definition video of this going to a central point. So the thing is, again, the enabling thing here to use effectively the imagery is driven by technological advances. It also enables you to deal with things like privacy. Why? Because you don’t send all data to a central repository and store it if it has no significance. And you can have a system decide that remotely, where it’s just stored locally and erased so that you don’t violate privacy. I mean, these are very delicate issues. But again, you get the enablement to deal with them from the underlying science and technology. The problem we’re faced with universally, however, is data complexity, and the velocity at which it’s arriving is now overwhelming us. We have built these systems of systems, the cities of today, at such complexity and so interwoven that a human cannot just stare at the data and say, ah, here’s the issue. That doesn’t work anymore. It didn’t work for something as simple as a child sitting in a bassinet. Now try to figure out how you make a city the size of Singapore work perfectly at all points of intersection between different areas, where you have mobility issues. You have traffic issues. You have education. You have all these things interact, because you have to get people to the schools and so on and so forth. This is where you really have to start to rely upon some other strategy. And we call it the emergence of the cognitive era. This is where you have man and machine collaboration. Because trying to do things independently and the complexity of this– it doesn’t work now. People love to call this AI. Forgive me, I like to call it accessible intelligence. Now there’s a long series of reasons I don’t like to use the term AI, the primary of which is what I call baggage. You see, when you say AI to the typical person, what they think of is Terminator, the movies where things come back and try to eat your head. This is not good. If you’re older, like I am, you think of Stanley Kubrick’s 2001 a Space Odyssey, with its very sweet voice of the AI, HAL 9,000. Which by the way, you do all know this, that they asked to name the machine an IBM 9,000. When we said what does it do, and they said, well, it kills everybody, we said no, you can’t do that. So instead they called it the HAL 9000, which if you move every letter one to the right, H becomes I, A becomes B, and L becomes M, and you’ve got IBM 9,000 anyway. Yeah, thank you. [LAUGHTER] Bet you didn’t know that one. The point is, this machine says, good morning, Dave, as it’s killing everybody he knows and tries to kill him. This is not AI as the way we want to think of it. Even more recently, Microsoft did this kind of cool experiment, which might not have worked quite the way they wanted, which you may or may not have read about. Anybody hear the story of TAI, T-A-I? They thought they had a really cool generalized AI system that could train itself. So they hooked it up to Twitter and said, learn! In 24 hours, they had a sex-crazed Nazi which they had to kill, because it was really getting embarrassing. We’re not quite ready for generalized AI, and this is nothing like it. But accessible intelligence basically extends the human capability so dramatically that it’s of enormous value. Now we’re not going to try to reinvent it. We call it AI, but I’m just cautioning, this ain’t TAI. It has nothing to do with it. Now human beings, we bring compassion. We bring intuition, our design skills. We bring value judgments. Some of us even bring common sense, though I have seen examples to the contrary, but we won’t talk about them. When you take a machine and you add things like the AI capabilities, you get deep learning, the ability to do discovery of incredibly subtle second, third, fourth order effects that nonetheless dominate. You can do large scale math, fact checking. Lord knows that’s needed nowadays. And the real kicker is when you have human and machine working together. That is when you get really remarkable outcomes. And this is essentially how you’ve taken something like a new capability, and you can apply it to coming up with new sustainable solutions that are extraordinary in many fields. I’m just going to show one as an example in medicine. This wasn’t done by us. This is actually a Bloomberg video, which I had to pay dearly for the right to use. But it makes the point really, really well. [VIDEO PLAYBACK] – Watson? – Who is David McCullough? – Correct. – IBM’s Watson, the technology that beat two humans on Jeopardy two years ago, may be the technology that helps doctors beat cancer. Dr. Arthur Forni at Westmed Medical Group in New York, is testing Watson in the form of an iPad app during its pilot phase. So, here Watson was given a very unusual case, a young Asian woman, non-smoker who has lung cancer. – So now you tell Watson to do its thing and it’s now looking at the data, you inputted the data from her imaging, from her biopsy. – In 17 seconds, Watson has analyzed about 3,500 textbooks and 400,000 other pieces of data. – This is the screen where it shows you that after looking at what you have so far, it would consider that you would do several other things for the patient that it doesn’t see yet in the chart. – In an instant, Watson has figured out what caused this cancer. – It turns out she has a very rare genetic mutation which is accounting for her cancer. – And offered three treatment options. – It has shown that treatment one has a 95% confidence of being effective, whereas treatment two is only half that. – Is this so obscure that you would have needed Watson for the diagnosis and treatment, or would an oncologist come up with the same treatment plan? – In fact, this is such a new discovery of this particular mutation– and it’s a relatively new drug as well– that my guess is that many oncologists may not be quite up to speed yet. [END PLAYBACK] BERNARD MEYERSON: And that’s the point. The human plus the machine give a value vastly greater than the sum of the parts. And by the way, people say, oh my god, it took all these human jobs. I would point out that Watson used 250,000 articles in coming up with this answer, all of which were written by innovative people who basically had done the basic research and development of the treatments, studying the genome. What it can do is it scales. Humans don’t scale, except when you come to Singapore and eat ridiculous amounts of food. But leaving that aside, the fact of the matter is that human beings don’t scale. You’re not going to be reading a quarter of a million articles in 17 seconds. Ain’t going to happen. I don’t care how good you are. And the fact is, a typical physician gets to read about five, and that’s not a criticism. And that’s a month, by the way. The reason is they’re busy saving people. This does the work that otherwise they wouldn’t have a hope in hell of doing, at a pace there is no chance they could pull it off. And this is part of sustainability, because at the complexity we’re not working, you need this capability. Now there are other disruptors out there that actually have implications on sustainability. And one of the huge disruptors is what we call Industry 4.0, which you can think of as the application of AI, machine learning, and similar technologies to different fields. But how they’re used is important. Think about how different the world would be if you can so accurately predict what’s going to happen that you can alter the future. It sounds like science fiction. But I want to show you something that actually is a clip that’s real. I mean, forgive me, it’s a commercial dramatization. All I can tell you is there’s a repairman who comes in the door here. He’s from KONE Elevator. This is actually done with that company. And let me just play it briefly, and then I’ll explain it. [VIDEO PLAYBACK] – Hey. – Pass, please. – I’m here to fix the elevator. – Nothing’s wrong with the elevator. – Right. – But you want to fix it? – Right. – So who sent you? – New guy. – What new guy? – Watson. [ELEVATOR PINGS] – My analysis of sensor and maintenance data indicates elevator three will malfunction in two days. – There you go. – You still need a pass. [ELEVATOR PINGS] [MUSIC PLAYING] [END PLAYBACK] BERNARD MEYERSON: If you think about it, we have for all of history talked about managing risk. It is a very different thing to be able to systematically eliminate it. This isn’t made up. It turns out that, as we’ve discovered– and this shouldn’t be a shock– if you listen to every moan and groan in a piece of equipment, you run it through a Fourier transform and turn the moans and the groans into frequency components, and then you compare that against the historical failures of those systems, guess what? It turns out that the appearance of certain frequencies correlate with the appearance of certain failures that are known. So you can actually send the repair person days or weeks ahead to a site with a specific part. Because you know just from the groaning coming out of the elevator what’s going to break. Now this wasn’t done once. This was done in an entire airport. And we actually do this in real time. We predict the future and alter it. And that’s a remarkable thing for sustainability. Because at the complexity of a modern city, if you don’t do this, the consequences are dire. You really do want to know if the elevators are going to quit in a building that’s got 57 floors that many of us are staying in right now. Because it’s a hell of a walk to the top. Now this isn’t something we’ve just applied in one place. We’ve applied this all over the world. And we’re not the only ones doing this, for obvious reasons. This is a necessity. The one on the bottom right is sewer systems. They have massive pumps. You really don’t want the sewer pumps to fail, especially if you’re living below sea level. That would just be unconscionable. Similarly, there are electrical generation plants, there are jet engines. You probably do want to know that jet engine’s going to quit before it actually does in flight, or even worse, during takeoff. On the top right, this guy’s got these yellow things in his hand. That yellow thing is actually a wrench used by a major aircraft manufacturer to screw the wings onto airplanes. You probably want to know when it’s going to drift low in torque and the wings are loose. Because you really don’t want the wings to fall off. By the way, that’s a real example. They actually had a whole bunch of these things drift out of spec. And they had to call back the airplanes and literally take out the guts to go screw the wings on tighter. This is really expensive. And to their great credit, they put in a smart system that predicts when these things will drift out of spec before they actually do. So these are incredibly powerful techniques. On the mundane side, in the middle, there’s a washing machine. One of the things we found is just from the noises this thing makes during its final test, we can predict which ones would basically in the first 20 days do you a favor and not just wash your clothing but the entire floor of your house as they unloaded their contents there. It’s really remarkable how much you can do with this kind of knowledge. But of course, you can think of ways of applying it in mission-critical aspects of sustaining your city. Now there are a bunch of real time examples that you can make, for instance, power outages. Can you imagine the consequences– you think there’s a bit of dependency there on electricity, judging by the neon lights throughout the city? It’s a huge sustainability issue for health care, public transport, heating, cooling, and failure is not an option. Having said that, take a look at what’s going on in Puerto Rico and the United States nine months after the last hurricane. There are sections of that segment of our society that do not have power back yet. Not only that, but the cost of mobilizing people to go and fix the problem goes up exponentially if you wait after the fact. Because once the damage has been done, the roads will be blocked. You can’t even get to the damage to make the repairs. Imagine for a minute that you minimize societal disruptions by using these technologies, AI, machine learning, to correlate, as on the left side, all of the weather companies’ very, very fine-grained weather data predictions, where literally we predict the weather within a square kilometer very accurately. You take that run that against all the historical data on outages, and you actually come up with a basic classifier that says, you know what? There’s a storm coming. And then the map on the right– that little red thing there– that’s where you need to put your assets. I know the storm will hit the whole area, but the only place there’ll be severe damage is right there. And make sure your trucks are at that location. Because even if the roads are blocked, they will have the correct equipment and people to make the repairs and bring the grid back up. That, in fact, is what we’re doing today. This isn’t future. And again, it’s just part of sustainability. But you would not believe the mathematical complexity done at the left side in terms of actually automatically generating the knowledge you need from the data. Now AI and IoT, internet of things and artificial intelligence, or accessible intelligence– people tend to think of them as sort of very high-end things, and we’re going to we’re going to go off and save cities and all. They have a lot of other applications. And I want to show you one that’s sort of near and dear. Every once in a while– one of the perks of being– my formal title at IBM is I’m the chief innovation officer. I tend to more appropriately use the term head geek. Now one of the reasons that provides me with air cover is geeks are not supposed to understand financials. That’s not true, actually, I run several multibillion dollar businesses. I do understand these things. But what it does allow me to do is sometimes make investments that I know damn well will never make us any money. But you know what? Sometimes you just do the right thing and worry about getting permission later. Now in deference to IBM, to be honest, when we did this work, instead of yelling at us for spending a ton of money for no benefit– financial, anyway– actually, we got lots of cheering. It’s part of our culture. So it’s something we did that’s a little unusual. And I will have to admit, it was a little humbling. I’ll explain later why. But let me just show you the work that was done using a combination of AI, machine learning, internet of things, in an environment one wouldn’t normally associate with doing so. [VIDEO PLAYBACK] [MUSIC PLAYING] – The rhino’s become one of the key species that is becoming endangered due to poaching throughout Africa, but now especially in South Africa. Up until now, poachers have been increasing in numbers. And they’ve become more militarized with weapons. So of course, we’ve had to do the same. This is not sustainable. The only way to do this better is to bring in technology and things that they do not have. – So the first thing people wonder is why are you collaring animals that aren’t rhino? The wildlife protection program aims to protect rhino populations by using mega herbivores as essential to detect poaching activity and within a reserve. A poacher enters an area, he’s more likely to come across an impala than a rhino. And so these animals will then behave differently. And as soon as there’s a change in their behavior, we’ll be able to detect where exactly a poacher is when he entered, and where he’s intending on going. – So these sentinel animals have collars with a sensor on them. The sensor transmits information across a LoRa network, which is a low power wide area network. – What we’re looking at is the sensor collecting the data on the animal that then gets back to the 3G network onto the IBM i2 platform. That data is then analyzed real time for the specific patterns we’re looking for. – Because the project works with IBM, MTN, as well as Wageningen University, each person has a different way of accessing this information, depending on the use of that information. So I find it useful to know where the animals are so I can plan my experiments. – And all of that can be referred to as the latest evolution in the use of technology called the internet of things. [END PLAYBACK] BERNARD MEYERSON: Preventing the extinction of an entire species through the use of AI, machine learning, internet of things, is really kind of cool. And what happened here was somewhat humiliating. Because of course, when I met these guys, my idea was, well, let’s put collars on the rhinos. And they said, great, when they fall over and die after they’ve been shot, we’ll know they’re dead. That’s not the idea. They pointed out, what you do is you put the collars, of course, as you heard, on gazelles and other very nervous animals. Because when the poachers come on to the area of the preserve, those animals will get spooked and run. Here’s the problem. They also get spooked and run when something’s trying to eat them, like a lion. Well, if you take the data of their movement and you run it through an analytics system, it can very quickly tell the difference between their behavior in the presence of a poacher versus in the presence of a lion. They behave really differently. They run with a lot more gusto when they’re being threatened with being eaten. And what it enables us to do is locate the poachers within a matter of meters. And then of course, they send a helicopter gunship out to strongly suggest the poachers sit on the ground and wait for somebody to go and pick them up. And only once did the poachers ever fire at the gunship. That ends really badly. And then word gets around that’s not a good idea, and it stops. Because the idea here is you don’t want to kill any people. You don’t want to kill any rhinos. And I should point out– why did we do this and why is it so vital? The UN keeps track of this. Over 1,000 rangers have been murdered by poachers as of 2013 or ’14. 1,000. This isn’t just about animals, this is about humans. And it is about sustainability. And it’s a war, and you want to fight the war where nobody gets killed. That would be a really good war. And so this is one of the ways of going about it. And as I said, we’re never going to make any money from it. But I will tell you that inside the company, we looked at this, said, yeah, this is a good one. We’ll just suck it up. Now innovations in technology find a lot of different ways to be applied that result in a sustainability outcome. Now one of the curious things is– one of the innovations we have, we’ve applied against hyperscale fraud. And you obviously ask, what does that have to do with sustainability? It’s kind of a simple answer. Right now, hyperscale fraud, fraud on a massive level, is enough to disrupt, of course, an individual business, or take down an entire nation. If you don’t think so, go to Iceland, or go to the UK and ask what happened to Ice Bank. You ever try to do business in a country that’s bankrupt? That’s exciting. The point is hyperscale fraud is something that you have to recognize, and you have to deal with. And the only way to steal $3.7 trillion dollars– trillion– that’s 5% of the world economy is stolen through fraud. How’s that for a number? And it’s not uniformly distributed, in which case, it would just be 5% loss. There are a whole bunch of companies that cease to exist every day because of this. The reason people get away with it is they can hide their detection by basically cooking the books. They basically alter their records. And by the time you figure out who has stolen your money, it’s long gone. What’s interesting is there’s a way of defeating it. On the left is the traditional way in which record keeping is done. Each party’s got their own records, and they share the records with the other party. You don’t know what’s true or false. Somebody says, I sent you 1,000 bananas. And the guy on the right who receives them says, OK, 1,000 bananas. But they’re only 970 in the trunk. They don’t know. Somewheres along the way, they found out– wait a minute, we’re 30 bananas short. And people throw their hands up and say, ah, what the hell. But you know what? 30 bananas– if you’re shipping 10,000 a day– the percentage there, you’re making a lot of extra money by cheating. That’s fraud. What’s done on the right is something called blockchain. And the very simple explanation of blockchain is there is a single ledger. Everybody doesn’t have books. There’s only one set of books. It’s called a hyperledger. What’s done is all transactions are sequentially deposited in a ledger. Now if everybody’s got a copy of the ledger, well, everybody can see everybody else’s data. The answer is no. Because blockchain works because you encrypt the data so that any individual company that’s on that same chain– there could be 500– there are 500 identical copies of that ledger held by these 500 companies. But each company only has the decryption keys to those parts of the ledger they’re entitled to see. The good news is if their government regulator is involved, they have decryption keys of everything. Because they need to see a transparent ledger. Now here’s the really good news. You’re trying to steal a vast sum of money, so you alter your ledger. The other 499 start screaming, he’s committing fraud. You can alter yours, but there are another 499 copies out there. What are you going to do about those? Conceptually, I’m simplifying it greatly. But conceptually, it’s very straightforward. And it is incredibly frustrating to those who are trying to commit fraud. The other thing it enables you to do in a complex system, a highly distributed system, is to track all activities. Because there’s one source of truth. Why is that important? Well, one of the challenges in sustainability, as you guys know, is food systems, right? Walmart, which is a huge retailer in the US and elsewhere, has an enormous basically supply chain for food. You all know historical evidence. Companies like Chipotle– it’s a restaurant chain in the US– they basically got a supply of what they believe was lettuce tainted with E. coli. And they were making people desperately sick, and they could never track down where it came from. It is your worst nightmare as a retailer to have this kind of a problem, where in a food chain somewheres, there has been a contaminated element introduced, and you simply don’t know how to find it, or you can’t find it fast enough to prevent deaths. Walmart, because it has this huge effort, said, you know what? We do track food. I mean, we keep records. But of course, in these highly distributed networks, it’s very hard to keep a record of everything right from the farm all the way to the table. Well, what we agreed to do with them is create a trusted food system where we had one ledger. We basically created a hyperledger very simple to use. The farmer literally, when they were putting something in a truck, pressed a button on their phone, said, I’m sending you, in this case, mangoes. And it prints out a label locally for him to put on the box that says, OK, this is Farmer Joe’s mangoes. And all along the way, everywheres this crate goes, it’s tracked and put into this single ledger. So everybody everywhere has a copy of what’s gone on. We then said, all right, what we’re going to agree to do is we’ll track mangoes by the traditional way they work and using a hyperledger. And at some point, somebody inside a Walmart is going to walk in room and say, guys, I am declaring this mango– literally, the one in my hand– is contaminated with salmonella. Tell me where it came from so I can take all of them off the shelves. Because the only other alternative would be to throw away millions of dollars of mangoes in the entire company supply chain. Now that was done. And here’s the outcome. To their great credit– god knows this is not a criticism– they were incredibly forthcoming in helping us test this. They did their own methodology, which is traditional. And it’s not their fault. This was designed by their supply chain. It took them six days, 18 hours, and 26 minutes, and they did find where the mango came from. Our guys went off in another room, pressed a button, and 2.2 seconds later, said, here’s the mango. 2.2 seconds. And that’s why you do it this way. Not only that, you can’t fake it. That data has absolute provenance. Because if you alter your copy of the ledger, every other copy that’s legitimate starts screaming that you have committed fraud. So this is an incredibly powerful tool. Because by the way, people are cheating and taking parts from crashed airplanes, literally, saying these are new. They polish them up and they then put them in an airplane. And you wonder why that rotor failed. Well, that rotor may have failed because it already had 8,000 hours on it, and somebody said no, it’s got no hours. They salvaged it and made it look pretty. This sort of thing– you need provenance in a complicated supply chain, like running a massive enterprise like a nation, state, or a city. So to sum this up, sustainability in this cognitive area is really kind of a cool thing. Governments literally can get near-instantaneous insights as to what’s happening and what’s possible to best serve society. Because man-machine collaboration becomes the norm to support it. However, to achieve the scale that we need to make this work, you have to store knowledge. You just can’t store raw data. That’s not going to work. Science will continue to deliver the tools required to craft a sustainable future. That’s a given. Science and technology underlie virtually all this progress. What does have to happen, though, is education is going to have to transform. Because a new generation of leaders are going to have to adapt to the reality that simply picking up a phone at a call center is no longer going to be a viable career. That will be done by machine. Because the fact is, it will be more accurate faster. But it’s not an innovative effort. This is something where we really have to look, essentially, much harder at our education system to ensure we have people who can populate what we sometimes call new collar jobs. These are jobs that are created whenever you have an Industrial Revolution of the sort that’s underway. The potential for positive impact here is absolutely limitless. It’s limited mostly by your imagination. The danger, of course, is you have to look at everything you do. Because the subtleties sometimes can actually undercut the benefit. As I said at the very beginning, if you have a hydrogen-powered car, but making the hydrogen produces more CO2 than you eliminate through that operation of the vehicle, you lose. So I don’t mean to make this easy. But if you do this rigorously and with some smarts, it can be an incredibly powerful technique. So thank you very much. And now is your chance to heckle. [APPLAUSE] MARIETTE DiCHRISTINA: Thank you, Bernie. So thanks very much. I thought that was an really excellent set of thoughts for us, and this idea of science delivering the tools, and it’ll be up to us to take best advantage of these opportunities. So that seems like a really great invitation to potential questions about how to take advantage of these opportunities. AUDIENCE 1: Hi, Dr. Meyerson. An amazing lecture, as usual, eye-opener. Yeah. I would like to actually point out something, and that is, you mentioned that Singapore has no natural resources. That relates to later innovation. And I was just wondering– because we have over 397 bird species, which makes up 4% of the world total. We have 255 hard coral species. That’s one-third of the world population. And I just could go on on and on. So why is that innovation always states that number one, Singapore has no natural resources, and second, if you talk about sustainability, learning from biodiversity is probably the best way. Because they managed to survive over the years. And there is so much things that we can learn using biomimicry. Perhaps we can use that for innovation. Thank you. BERNARD MEYERSON: There are about eight questions in there, so let me just try the first one. I beg to differ. You understand that the conventional definition of a resource is a consumable. And Lord knows, please don’t decide to eat the birds. [LAUGHTER] I beg you not to do that. Biodiversity in Singapore is huge. But in the classical sense of consumables, water, energy sources like oil and all of that, Singapore isn’t rich in those. It has, however, a tremendous wealth in its people. That is a huge natural resource, which basically, has managed to turn it into this astonishing society. But I’m not going to argue with you over whether or not we want to count the birds. Because please, like I said, don’t eat them. You will notice that I intentionally actually avoided– the real point here was about how the current mainstream science and technology are enabling all sorts of new ways of attacking sustainability. By no means am I ignoring the fact that biodiversity and all the other facets of biology also play a role in this. That was part of the reason we had Chris [INAUDIBLE] in there at the beginning. Again, it’s a bleeding edge aspect to it. Because biodiversity has been known as an asset for ages, whereas CRISPR-cas is only just now coming into being. And in fact, I think they only approved the use of it last year for human experimentation. It’s brand new. So I was gating it by time, not ignoring it. It’s just we’re talking about literally the most recent evolutions of technology and science that occurred literally now. I mean, we’re talking within this year. There’s no question there’s enormous amount of history that is out there. But this audience would be very familiar with it, or at least I would assume so. That’s why we didn’t get into it. Also, I only have an hour, and that takes about 18 if I actually wanted to cover all of that. AUDIENCE 2: Sasha [INAUDIBLE] from NTU. This is all nice. Because you showed us nice examples where in a highly technologically developed society, things can be made better. But we have a world where that part is small. And the biggest part are people who have no affordable access to all what we are saying here. And the problem is that the science which is developed kind of in the developed world doesn’t find its way where it has actually to be in a given time. I give you one example. The golden rice has been developed in the ’80s. Not yet with CRISPR, but we know exactly where the Vitamin A is in the rice and so on. That would actually help one third of the world population to keep their eyesight and their eyes in good shape. Because they are all low in Vitamin A. The same is actually with iron, also. But take the rice. It was around at the end of the ’80s. It’s still not in the fields. And the problem is somewhere totally different. It’s a societal problem. It’s a political problem. Because these things have to go through decision making in a political system where– you asked who believes in global change? Well, I don’t really believe in global change. I believe in the facts which I see, or I understand the facts and I can do that. And the problem is that many of these things go down to a belief where we scientists have almost no access to bringing that in quote, “belief” away, and that people are actually looking at the facts. And many of the things which we develop here in the technological societies look very great and all these kind of things, have a too-long lag phase to get there where it’s actually needed. BERNARD MEYERSON: So let me again– so I took out about 20 slides that I have, one of which I was going to pull up, but it’s easier to talk to. In fact, what you should have asked me as why do you call it accessible intelligence? The reason is an answer to your question. The reason we call it accessible intelligence is a farmer in sub-Saharan Africa does have technology. Astonishingly, low-cost handsets are almost pervasive now in sub-Saharan Africa. They’re not the $1,000 smartphones. They’re the $43 phones using SD-TDMA from China, but it doesn’t matter. The whole point of that is this intelligence is all web-based. You don’t build any of this locally. And the vision that we have, to be blunt, is exactly to address your issue. The reason we call it accessible intelligence is this farmer should literally just be able to dial *6, that’s it, and ask a question in natural language, in their own language, that says, what should I plant so my family doesn’t starve? And literally, the system will look at every available source from satellite imagery of ground conditions, what are the history of the vegetation in the area to make sure you don’t repeatedly plant crops that deplete, and literally instantaneously say you should plant golden rice, or sorghum, wheat, or whatever. This is the time you should plant it under these conditions for that exact reason. Because the dissemination– I actually have a chart that says it gets rid of the digital divide. The whole point of AI as we vision it is to address that issue. Because it bypasses the normal roadblocks you have by making it pervasively available, and essentially impossible to block. Now will that happen? It is beginning. But it is actually the whole point of calling it accessible as opposed to artificial. Not to mention the word artificial always bothers me. Because when you look it up, it says fake. This isn’t fake intelligence. It’s pretty damn bright. But you make a very valid point. And the whole idea of accessible intelligence is to address exactly the issue you raise. But that’s a whole ‘nother discussion, and an important one, honestly. Maybe I pulled the wrong chart out. But you are absolutely correct that otherwise, this is of no value. Because the world is not uniform, so what works in Singapore may not work in Sudan. Understood and agreed. NICK CAMPBELL: Hi, Bernie. So it’s Nick Campbell from Nature research. So I think one of the really interesting concepts you brought up in your talk was the idea of social sustainability as well, and not leaving people behind. And you also mentioned the idea of new collar work, right? And so one of the things that I really don’t see at the moment globally is a properly articulated vision from the politicians, the lawmakers, about how we turn blue collar into new collar. And I just wonder if you’d like to comment on that, and whether you see any political will out there? Because I mean, rather than saying, bring back coal, we want to have politicians who are saying, we’ve got a vision for turning your blue collar job into a new collar job. BERNARD MEYERSON: There actually– OK, so a couple of answers. It depends where. In the US, out of sheer frustration, we stepped up ourselves, ponied up our own money, and started a program called P-TECH. P-TECH actually intercepts kids who are from– most of the time, these are communities that are not doing well, to be blunt. But these are kids who want to excel. We intercept them at the end of junior high school. And there’s a six-year program where, in fact, you train for what we call the new collar jobs. And you come out with an associate degree– not just a high school degree, an associate degree– in fields germane to what we’ve been discussing here. And we funded it ourselves, and it has been wildly successful, to the point that there are over 100 P-TECH schools now all throughout the United States. Two days ago in Singapore, we just did the same thing. We announced with the Minister of Education two rooms over from here that program in Singapore. Now if you’re saying is it on a global basis– P-TECH is turning out to be interesting, because it is being adopted. The real challenge will be that the very places where you have the most exposure to the loss of these blue collar jobs right now, it isn’t yet gaining traction. So I couldn’t agree more of the necessity. And in fact, education– I just had it on a sub note, as opposed to a major topic– is the key, in fact, to getting this done. So do I see the will? Actually, yes. I’ve seen it in the US. I’m seeing it in a number of other countries. But am I seeing it on a global basis that would make me comfortable? Absolutely not. And it is a necessity. Because this is going to happen. Financially, if you can take literally, 5,000 people in a call center, who are hugely expensive– because they only stay on average, three to six months before they move on. If you can take those 5,000 and reduce it to 50 who only deal with level 3 queries that you just, for whatever reason, can’t handle, that means that you’re suddenly looking at 4,950 people who are out of work. And you better do something about it, or else there’ll be social unrest and all the ugly that comes with it. So I completely agree with your point. And answer, no, I do not see evidence yet of it being pervasive. But it is starting. AUDIENCE 3: So I’m Venus [INAUDIBLE] from UN NSW in Sydney. And I do want to thank you. That was really awesome and tremendous. And I think one of the things that we are doing at UN NSW as part of our ability to transform different types of waste resources, so whether we’re talking about electronic waste or glass or plastics. And the whole list goes on and on. And of course, that is something that touches every community globally. But let me just pick up on what we’re doing in Australia. And I’d like to sort of, in a way, capture the points you’ve made so beautifully, that how I can see the relevance to where the transformation of waste can play a role when you bring in all of those digital solutions you’re talking about, in terms of addressing the social challenges. And so part of that distributed manufacturing– and I notice you had that on your list– we’ve developed technologies that are all around micro-factories. And micro-factories are indeed all around, creating this distributed manufacturing. And so the critical part in all of that is that what people love about it– and indeed small communities, regional towns where the population might just be 1,000 people, where they could never imagine manufacturing jobs coming to that locality. But indeed, what they can see is the opportunity you’re bringing in. All of these jobs– distributed manufacturing, decentralized micro-factories– to be able to create that as an opportunity. And what I loved about what you said was in a way, with a lot of our technologies, the ability to actually allow people who don’t necessarily understand what that waste could well be. Because we know our products are complex. But they can actually now pick up on all of that data that we’re gathering through a lot of the visual techniques. Because part of our technique is looking at the waste. And you don’t necessarily have to know the answer as to what your micro-factory might be able to produce. But I can just imagine that using the kinds of things you’re talking about, we would be able to actually use some of those ideas and new concepts to spur on. And I wonder whether then the conversation should be a far more holistic one around how do we then bring in policy makers to be able to see that there is manufacturing that can happen at a local level, that truly is a value-add, both in the context of the social outcomes as well as economic outcomes. And I wonder whether there’s some thought, I guess, or comments you’d like to see where all of this could potentially have a massive impact? BERNARD MEYERSON: Look, you’re right. When you try to overcome– let me put it this way. We’ve actually gone, despite what you may think, into sub-Saharan Africa and elsewhere with these very technologies and had great success and got great acceptance. Unfortunately, it was when people’s hair was on fire, literally. During the Ebola crisis– we have a lab in Kenya. And people were just dying by droves in Freetown, Sierra Leone. The government– it was very intrusive. We had to take data on literally who’s dead where. Where is their blood? Where are there people who are not isolating themselves despite knowingly being infected, and so on and so forth. You had to get all this incredibly sensitive data. And you wanted to collect it and then be able to tell the government where they needed to send their resources to most effectively deal with this incredible crisis. People did not want to call you up and turn themselves in, or turn others in. I mean, it was an incredible crisis. But we managed in this environment, where their hair was on fire, to convince the government to allow us to skirt all kinds of privacy rules knowingly, I mean, openly. We actually– you’re supposed to, for instance, capture the name of somebody who calls up, and says, hey, there’s a non-compliant person next door who’s going to kill people. Instead, what we did was we actually put in a company, a third party, that anonymized all the SMS messages sent to us telling us this data. And they were not even allowed to keep the data. They literally took the data, sent it to us with no attribution. We took all that data to Kenya. We basically ran the analysis as to where basically, you would best deploy your resources. And it made a tremendous difference in crushing the Ebola crisis in Sierra Leone. So you can do it. But you’re on a key point. Getting the social will, many times, unfortunately, has taken a crisis to drive it, and that’s regrettable. If you’re asking, should it be done, could it be done? Absolutely. It’s not only of interest, it’s something that we could do. And we should talk later. Because the World Economic Forum actually has got a whole bunch of programs going in the Fourth Industrial Revolution, where you can almost treat that as one of the outcomes. And maybe it would fit under that banner. They have a center in San Francisco that might be willing to engage on that. I know the guys who run it. So there may be ways of doing it. But it is a challenge, because when you start dealing with governmental entities, unless there’s really a crisis situation, it’s very hard to get the velocity you’d like to see in the engagement. But it’s worth making the attempt, you know? You will never find out if it works without making a shot at doing it. So thank you very much. I know we’re out of time. And we can keep hacking at it. I’ll be around. Let’s go get a cup of coffee and take a break. Thank you. MARIETTE DiCHRISTINA: Thank you. Thank you so much, Bern. [APPLAUSE]

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *