WEBVTT - How artificial intelligence can help humans fight climate change | EP 52 0:00:00.090 --> 0:00:05.220 The following is a C and a podcast. This is 0:00:05.220 --> 0:00:08.550 the climate conversations, I'm Jamie Hill today. I'm talking about 0:00:08.550 --> 0:00:12.209 climate change and artificial intelligence. As far as a climate 0:00:12.210 --> 0:00:15.180 crisis goes, we're all looking for solutions, whether it's the 0:00:15.180 --> 0:00:18.330 pursuit of carbon neutrality or simply trying to push the 0:00:18.329 --> 0:00:22.169 boundaries of energy efficiency. So what can I add? Well, 0:00:22.170 --> 0:00:26.700 on human interventions right now face limitations from monitoring and 0:00:26.700 --> 0:00:31.380 forecasting to optimizing energy use in datacenters. Buildings and power 0:00:31.380 --> 0:00:35.580 grids could be a tool in fighting climate change. Yes, 0:00:35.580 --> 0:00:39.170 advances in deep learning and in computational power have made A.I. 0:00:39.210 --> 0:00:43.410 less science fiction, more non-fiction and here and now. But 0:00:43.409 --> 0:00:45.900 are there also downsides of using A.I. to deal with 0:00:45.900 --> 0:00:51.960 a problem as complex as climate change? With me today 0:00:51.960 --> 0:00:55.590 is Marcus Croft, professor of chemical engineering and director of 0:00:55.590 --> 0:01:00.690 Cambridge Care's the University of Cambridge's Research Centre based in Singapore. 0:01:00.900 --> 0:01:04.380 Professor Craft is an expert in using computational models to 0:01:04.380 --> 0:01:09.449 reduce carbon emissions. And his new book, Intelligent Decarbonisation, explains 0:01:09.450 --> 0:01:13.500 how I might be able to end climate change. Marcus, welcome. 0:01:13.709 --> 0:01:16.710 Hello, everyone. Thank you for giving me a chance to 0:01:16.709 --> 0:01:17.729 be on the podcast. 0:01:17.819 --> 0:01:19.800 Great to have you. I'm going to jump straight into 0:01:19.800 --> 0:01:23.580 the first question for the average man in the street. 0:01:23.610 --> 0:01:25.350 I'll go out on a limb, and I'll say that 0:01:25.350 --> 0:01:31.410 maybe artificial intelligence wouldn't be completely understood fully right? In general, 0:01:31.410 --> 0:01:34.770 we know it helps us analyze data and make predictions, 0:01:34.770 --> 0:01:38.700 optimize systems, perhaps before we get into climate change. Give 0:01:38.700 --> 0:01:41.309 us a quick sort of overview of how EIA is 0:01:41.310 --> 0:01:44.459 already being used today as part of our daily lives. 0:01:44.850 --> 0:01:47.880 Let me not you wouldn't be entirely aware of 0:01:48.240 --> 0:01:52.830 the use of the AI that impacts on everyone. Bailey 0:01:53.460 --> 0:01:58.620 is in the search engines and in customer feeds and 0:01:58.620 --> 0:02:03.900 web pages like Amazon or similar companies. This is where 0:02:04.200 --> 0:02:08.669 we benefit most of the AI right now. In particular, 0:02:09.120 --> 0:02:13.880 machine learning has been enormous for about three years ago, 0:02:13.889 --> 0:02:18.510 there has been a breakthrough in natural language processing using 0:02:18.510 --> 0:02:23.280 the Transformer machine learning networks. We have also increased our 0:02:23.280 --> 0:02:28.739 ability to do automatic translation, something I enjoy using almost 0:02:28.740 --> 0:02:32.580 every day and all to try to understand what we 0:02:32.580 --> 0:02:36.000 are looking for when we type a search request into Google, 0:02:36.000 --> 0:02:36.810 for example. 0:02:37.290 --> 0:02:39.940 So I'm going to jump on the term that you use, obviously, 0:02:39.950 --> 0:02:41.970 is something that we are quite aware of here in 0:02:41.970 --> 0:02:45.419 our newsroom. And as machine learning to sort of Segway 0:02:45.419 --> 0:02:48.180 into climate change, I'm sure there are machines out there 0:02:48.540 --> 0:02:52.530 which are learning about man's involvement and attempts to manage 0:02:52.530 --> 0:02:55.800 climate change. Give us a broad sense, then, how AI 0:02:55.800 --> 0:03:00.570 and technology in general has already supported our plans various 0:03:00.570 --> 0:03:03.690 forms to get to net zero mitigating the impacts of 0:03:03.690 --> 0:03:07.530 climate change. What kind of EIA is this and how 0:03:07.530 --> 0:03:11.220 might it apply to a specific area? See our water 0:03:11.220 --> 0:03:13.320 usage or electricity generation? 0:03:13.530 --> 0:03:17.040 This is a very good question, and it's hard to 0:03:17.040 --> 0:03:20.280 respond to. One of the problems with the response is 0:03:20.280 --> 0:03:24.060 that what do you mean when you see different people 0:03:24.060 --> 0:03:28.410 mean different things when they use the word on the 0:03:28.770 --> 0:03:32.370 and sort of thing, something magic happens and you get 0:03:32.400 --> 0:03:35.760 a fantastic solution out of nowhere. And I don't think 0:03:35.940 --> 0:03:39.480 this is actually the case if you look at today's technology. 0:03:39.900 --> 0:03:44.100 Of course, there is a trivial way to reduce CO2 emissions, 0:03:44.100 --> 0:03:47.730 and that is just to stop using fossil fuels. But 0:03:47.730 --> 0:03:52.020 that would have enormous impact on our society and is 0:03:52.020 --> 0:03:56.760 generally regarded as not acceptable for staff. Typekit very difficult 0:03:56.760 --> 0:04:00.090 to go back to Stone Age. So, OK, what can 0:04:00.090 --> 0:04:03.120 we do instead? It is clear that we have to 0:04:03.120 --> 0:04:07.710 replace the fossil fuels with something else, and depending on 0:04:07.710 --> 0:04:11.190 which sector you're looking at, different solutions are coming out. 0:04:11.490 --> 0:04:15.390 The more simple one was actually in the electrical power 0:04:15.390 --> 0:04:19.740 networks here. Many people may have heard of the term 0:04:19.740 --> 0:04:25.650 smart grid, so this is the first example where digitalisation, 0:04:25.770 --> 0:04:29.910 not just A.I., is changing the way we operate. This 0:04:29.910 --> 0:04:35.090 is partly necessary because if we use different energy sources. So, 0:04:35.100 --> 0:04:39.239 for example, the sun or wind power, then we have 0:04:39.240 --> 0:04:43.980 much more fluctuations. The on the grid does not necessarily 0:04:43.980 --> 0:04:48.570 match the input of energy that we have. So we 0:04:48.570 --> 0:04:53.219 have to look into, for example, clever storage technology in 0:04:53.220 --> 0:04:58.020 all these aspects, which would be handled by mathematical algorithms. 0:04:58.260 --> 0:05:02.370 And here we go. What do these algorithms look like today? 0:05:02.520 --> 0:05:05.670 In many cases, because we have so much data, we 0:05:05.670 --> 0:05:08.489 actually don't need to sit down and do physics. But 0:05:08.490 --> 0:05:12.090 we could use just the data and sometimes and very 0:05:12.089 --> 0:05:15.570 often the a combination of two. And that helps us 0:05:15.779 --> 0:05:19.710 control the smart grid or make the smart grid smart, 0:05:20.070 --> 0:05:25.980 which is then helping to secure the electricity issue for 0:05:26.160 --> 0:05:30.810 the population and industry. I will come into the control 0:05:30.930 --> 0:05:31.980 of smart grids, 0:05:32.610 --> 0:05:35.250 so I'm going to use the example of dual island 0:05:35.279 --> 0:05:38.010 here in Singapore to drill down a little bit into 0:05:38.010 --> 0:05:41.580 how grids are smart, how energy systems are smart, and 0:05:41.580 --> 0:05:44.400 maybe I will help them get even smarter, right? In 0:05:44.400 --> 0:05:47.550 your book, you've written about use cases that are associated 0:05:47.550 --> 0:05:50.890 with your island. What did your research? Find on how 0:05:50.890 --> 0:05:54.190 iOS could do things from reducing costs to emissions in 0:05:54.190 --> 0:05:58.240 a major petrochemical hub like Drone Island and what other 0:05:58.240 --> 0:06:01.210 applications are being tested here and what holds the most 0:06:01.210 --> 0:06:02.470 promising in your view? 0:06:02.920 --> 0:06:05.030 I have to say that if you look at your 0:06:05.029 --> 0:06:08.800 own island, that is potentially the most difficult spot. So 0:06:09.220 --> 0:06:12.890 for variety of reasons, I personally believe that you will 0:06:12.890 --> 0:06:16.480 try to do things in other areas, in particular in 0:06:16.480 --> 0:06:20.620 the smart city projects much earlier. So you have basically 0:06:20.630 --> 0:06:23.680 two problems. The first problem is, of course, to supply 0:06:23.680 --> 0:06:28.120 the energy in a carbon footprint free way. At the moment, 0:06:28.120 --> 0:06:32.500 the energy is by meaning a gas power stations, so 0:06:32.529 --> 0:06:37.779 you would have to selectively, but which the company owned. 0:06:37.779 --> 0:06:42.520 You're allowing to their own proprietary power stations. And we 0:06:42.520 --> 0:06:47.580 have a lot of oil generators just for safety. So 0:06:47.650 --> 0:06:52.719 and backup replacing the energy will not be straightforward, although 0:06:52.930 --> 0:06:55.840 you could argue that hydrogen may be the way forward. 0:06:56.440 --> 0:06:59.860 The second problem you have with the chemical industry is 0:07:00.130 --> 0:07:04.719 that the product itself have a high carbon footprint in 0:07:04.720 --> 0:07:08.799 terms of logistics transporting them around, and b they are 0:07:08.800 --> 0:07:12.430 made out of fossil fuels, you know, so that the 0:07:12.430 --> 0:07:16.240 refineries produce the commodity chemicals that are then going to 0:07:16.240 --> 0:07:19.720 our product that needs to be replaced. We have a 0:07:19.720 --> 0:07:23.620 double challenge, so to say. So if you can solve 0:07:23.710 --> 0:07:26.980 the problem with two or island, you're basically on top 0:07:26.980 --> 0:07:31.070 of things, which is very exciting. And that's partly why 0:07:31.070 --> 0:07:34.420 I thought we should look at that. But clearly, I 0:07:34.420 --> 0:07:36.490 would be lying if I said, we have a solution 0:07:36.790 --> 0:07:42.010 next Monday. Okay, this is long term research with particular 0:07:42.010 --> 0:07:47.290 things you can do now. For example, you use overall 0:07:47.290 --> 0:07:52.720 network of companies under alignment. So, for example, Company one 0:07:52.960 --> 0:07:56.980 may produce waste heat, which then can be used by 0:07:56.980 --> 0:08:02.430 another company. You see cost energy resources by doing so, 0:08:02.440 --> 0:08:05.080 and we do indeed have the study where we looked 0:08:05.080 --> 0:08:09.250 into this. We also did a little study just to 0:08:09.310 --> 0:08:12.660 highlight how difficult the problem is where we say if 0:08:12.670 --> 0:08:16.120 you want to be brought to the extreme place nuclear 0:08:16.240 --> 0:08:20.560 reactors under a. There are many reasons why you wouldn't 0:08:20.560 --> 0:08:23.020 want to do that. But if you were just looking 0:08:23.020 --> 0:08:27.280 at the CO2 footprint, then this could be a way for. 0:08:27.370 --> 0:08:32.230 But it was more an academic study in order to 0:08:32.230 --> 0:08:36.429 highlight the consequences if you really want to follow through 0:08:36.790 --> 0:08:38.770 with zero carbon footprint challenge. 0:08:39.700 --> 0:08:42.069 I would imagine, therefore you would sound as if in 0:08:42.070 --> 0:08:46.240 the short term, the very short term things that maybe 0:08:46.240 --> 0:08:50.770 can be improved on within a very unique situation like 0:08:50.770 --> 0:08:53.920 dual island would be improvements on the fringe rather than, 0:08:54.100 --> 0:08:57.339 as you see really huge structural changes. And on the fringe, 0:08:57.340 --> 0:09:00.970 you are talking about efficiencies in energy usage. How far 0:09:00.970 --> 0:09:03.550 do you think the eye has already helped there and 0:09:03.550 --> 0:09:06.220 what more is out there in terms of research that 0:09:06.220 --> 0:09:09.939 can actually push the boundaries of energy efficiency in really 0:09:09.940 --> 0:09:12.550 intensive environments like Jurong Island? 0:09:13.059 --> 0:09:14.800 I have two things to say to that. If you 0:09:14.800 --> 0:09:17.590 look at your items, you will see most of the companies. 0:09:17.590 --> 0:09:23.740 There are big international companies. So, for example, Exxon being 0:09:23.740 --> 0:09:28.140 one of them or both aid and others, they have 0:09:28.150 --> 0:09:32.020 fantastic research labs. They know very well what are the 0:09:32.020 --> 0:09:38.500 newest methodologies be on the forefront? Developing this methodology so 0:09:38.500 --> 0:09:41.770 they know about. But equally, you have to realize that 0:09:41.890 --> 0:09:46.060 the investment cycle in such big plant actually quite long. 0:09:46.540 --> 0:09:49.900 There is always a balance between sort of capex and opex, 0:09:50.170 --> 0:09:54.790 and that will lead to an additional time delay, although 0:09:54.940 --> 0:10:00.740 sometimes technology may already be known. Implementing them, making the investment, 0:10:01.120 --> 0:10:05.380 getting a return on base, that is what needs to 0:10:05.380 --> 0:10:07.810 be looked at if you want to estimate the time 0:10:07.809 --> 0:10:10.900 it takes to completely change the industry. In my view, 0:10:10.900 --> 0:10:14.679 it also becomes clear that one of the problems is 0:10:14.679 --> 0:10:18.370 actually policy making because policy making will have a direct 0:10:18.370 --> 0:10:21.849 impact on these costs. Most countries are aware of it, 0:10:21.850 --> 0:10:25.240 and if you look to Europe, they discuss carbon tax 0:10:25.450 --> 0:10:28.960 and trading. And I think over the next five to 0:10:28.960 --> 0:10:33.340 10 years, it become increasingly the method of choice to 0:10:33.340 --> 0:10:37.390 push new technology in to make this investment cycle shorter. 0:10:42.900 --> 0:10:46.600 I'm going to switch accommodation to another large company, Google. 0:10:46.740 --> 0:10:50.579 And you make reference to them in your book Intelligent Decarbonization, 0:10:50.580 --> 0:10:55.439 where you mention an Irish system called DeepMind, helps cool 0:10:55.470 --> 0:10:59.130 Google's data centers, and they are obviously quite big users 0:10:59.580 --> 0:11:04.410 of energy. And it's apparently helped cut energy consumption by 0:11:04.410 --> 0:11:08.070 30 percent, it said. Explain how that worked. Will this 0:11:08.309 --> 0:11:12.060 technology be game changers in a way in terms of 0:11:12.059 --> 0:11:13.110 energy efficiency? 0:11:13.350 --> 0:11:16.229 It will definitely help. It is not a Magic Machine 0:11:16.230 --> 0:11:20.179 learning hospital using controlled strategies to depend on the reaches 0:11:20.190 --> 0:11:23.910 of your data. And I'm sure this could be rolled 0:11:23.910 --> 0:11:28.380 out of the areas where digitalisation has been developed fine. So, 0:11:28.380 --> 0:11:32.189 for example, last year I was trapped in my German 0:11:32.190 --> 0:11:35.550 hometown in London, and that's not Wow. Okay, what am 0:11:35.550 --> 0:11:38.100 I going to do? So I decided to contact the 0:11:38.100 --> 0:11:41.400 mayor and ask them thanks to a smart city. And 0:11:41.400 --> 0:11:44.790 you have to know that my little hometown, just 40000 0:11:44.790 --> 0:11:47.490 people there are really poor and they don't have money. 0:11:48.030 --> 0:11:50.819 And I said, Let's do smart city together. And as 0:11:50.820 --> 0:11:54.040 a consequence, they were actually quite open. And we have 0:11:54.120 --> 0:11:59.160 a project which basically optimised the distribution network and we 0:11:59.160 --> 0:12:04.800 could be using a variety of methodologies that will reduce 0:12:05.100 --> 0:12:08.130 the cost for them by 20 percent. And we could 0:12:08.130 --> 0:12:11.190 use the CO2 impact into what I'm trying to say 0:12:11.190 --> 0:12:14.770 here is what happened with Google. It's not an exemption, 0:12:14.800 --> 0:12:18.000 something that we see no doubt in many more areas. 0:12:18.420 --> 0:12:22.170 You said API is terribly expensive, and if you look 0:12:22.170 --> 0:12:25.439 at the big data centers, you have the carbon footprint, 0:12:25.440 --> 0:12:29.339 which would be high. And this is true. However, it's 0:12:29.340 --> 0:12:31.470 a bit like the electric car. If you look at 0:12:31.470 --> 0:12:34.410 the electric car, they think that solves every problem. But 0:12:34.410 --> 0:12:38.640 it's not all me if and only if the electricity 0:12:38.940 --> 0:12:42.510 that goes into either electric cars or in the beauty 0:12:42.510 --> 0:12:46.949 sector is carbon footprint free. So it's either from solar 0:12:47.429 --> 0:12:52.319 or from wind or from fusion within. So it's not 0:12:52.320 --> 0:12:54.679 a problem. And in fact, one can say, you know, 0:12:54.690 --> 0:12:57.550 we don't have the energy from such as plenty of 0:12:57.630 --> 0:13:01.500 energy and our haven't you know what we haven't really 0:13:01.500 --> 0:13:05.849 managed to do is to get our head around how 0:13:05.850 --> 0:13:09.550 to use solar in the right manner. And this is 0:13:09.570 --> 0:13:13.440 a few that has developed enormously quickly over the last 0:13:13.440 --> 0:13:17.250 few years are now talking about perovskite cells with very 0:13:17.250 --> 0:13:22.219 high efficiencies and here. This is also very important feed 0:13:22.230 --> 0:13:26.760 for air. So one of our activities is to automate 0:13:27.150 --> 0:13:30.150 chemical laboratory, but not just the laboratory or to automate 0:13:30.150 --> 0:13:34.740 the site. Right. So you have robots that perform experiments 0:13:34.740 --> 0:13:40.319 to find more sustainable synthesis rules for better materials, and 0:13:40.530 --> 0:13:44.430 we can use machine learning to pick the right materials 0:13:44.700 --> 0:13:48.150 that will have, in my view, enormous impact on the 0:13:48.150 --> 0:13:51.360 development and the efficiency of solar cells, which then in 0:13:51.360 --> 0:13:55.770 turn will be because everything will be connected, can then 0:13:56.160 --> 0:14:00.080 make computing center just the ones that will just carbon 0:14:00.179 --> 0:14:01.980 footprint free. And then off we go. 0:14:02.160 --> 0:14:04.640 That's sort of a related point that I would have 0:14:04.650 --> 0:14:08.809 then is related to the cost of the machine learning, right? 0:14:09.120 --> 0:14:14.380 Using DeepMind, as my example is, showed that just training this, 0:14:14.400 --> 0:14:18.540 the AI system is also energy intensive in and of itself. 0:14:18.780 --> 0:14:21.630 And there's been studies that find that training a huge 0:14:21.630 --> 0:14:26.310 system like that could consume significant amounts of energy and 0:14:26.310 --> 0:14:29.100 emissions as well. Is this sort of something that you 0:14:29.100 --> 0:14:31.560 keep in mind as you look at research in terms 0:14:31.560 --> 0:14:34.830 of how and whether there is a trade off between 0:14:35.100 --> 0:14:39.810 his own energy consumption and his purpose in optimizing energy use? 0:14:39.960 --> 0:14:42.660 I have three things to say to that. The first 0:14:42.660 --> 0:14:46.650 thing is that you are right when you say the 0:14:46.650 --> 0:14:51.450 training of an artificial neural network and thinking of particular 0:14:52.050 --> 0:14:56.070 about GBP three, look to whether you use a natural 0:14:56.070 --> 0:15:01.830 language network. It's also based on this principle technology that 0:15:02.070 --> 0:15:08.070 basically has used unprecedented amounts of data. In China, there 0:15:08.070 --> 0:15:12.000 is a similar network being trained, but you don't have 0:15:12.000 --> 0:15:16.200 to do it training every time. So you train such 0:15:16.200 --> 0:15:19.560 a network and then you have to do all minor 0:15:19.560 --> 0:15:24.690 modification to use it so that the benefit multiply. Eventually, 0:15:24.870 --> 0:15:28.800 if you distributed over all machine learning algorithms that are 0:15:28.800 --> 0:15:33.210 then based on something that could be or transform the network, 0:15:33.450 --> 0:15:38.340 then the energy use is actually not that dramatic. Second is, 0:15:38.820 --> 0:15:41.670 if you look at A.I., it would be a mistake 0:15:41.670 --> 0:15:44.830 to just. Weighted to the kind of structure of the 0:15:44.830 --> 0:15:48.740 neural network and the way we trained, this field is 0:15:48.740 --> 0:15:53.640 rapidly evolving, in my view. We are just about to 0:15:53.670 --> 0:15:58.620 ship a new revolution in the way we do this API. 0:15:59.210 --> 0:16:03.260 So far, these networks are based on data and data, 0:16:03.260 --> 0:16:06.310 and though there is no understanding, they're going to use it. 0:16:06.440 --> 0:16:09.110 But if you look at our brains, we have similar 0:16:09.110 --> 0:16:12.320 performance and use only a fraction of the energy and 0:16:12.320 --> 0:16:16.130 we can learn much, much faster, which is why and 0:16:16.230 --> 0:16:19.370 probability that we will get to grips with it is 0:16:19.370 --> 0:16:23.360 very high. So, for example, we spend a lot of 0:16:23.600 --> 0:16:27.860 effort to develop policies to develop the work model. In 0:16:27.860 --> 0:16:31.100 our book, I call it the word avatar, which is 0:16:31.100 --> 0:16:36.350 basically nothing else but a representation of the world in 0:16:36.350 --> 0:16:39.830 terms of knowledge in this type of space. Now, of course, 0:16:39.830 --> 0:16:43.880 you could not just learn on the data as such, 0:16:43.880 --> 0:16:47.240 but you learn on stuff that is already known. And that, 0:16:47.240 --> 0:16:50.720 of course, will have enormous impact. If you ask me, 0:16:50.990 --> 0:16:54.979 will the training for networks that can do similar things 0:16:54.980 --> 0:16:58.610 like geometry be always as expensive as it is now? 0:16:58.640 --> 0:17:01.430 I don't think so. I'm up set with the next 0:17:01.430 --> 0:17:05.120 few years. There will be significant progress in that respect. 0:17:05.359 --> 0:17:08.900 The final point I got already made it is our 0:17:08.900 --> 0:17:13.220 brain doesn't use that much energy, which basically tells you 0:17:13.820 --> 0:17:16.550 there is a lot of room for improvement. Frankly, we 0:17:16.550 --> 0:17:19.970 just haven't quite put our heads around it. I believe 0:17:20.180 --> 0:17:24.440 although I cannot pinpoint specific areas that will improve, I'm 0:17:24.440 --> 0:17:27.770 sure that because there is so much room for improvement 0:17:27.770 --> 0:17:28.940 that will happen. 0:17:35.250 --> 0:17:37.920 When people talk about the vast sort of potential that's 0:17:37.920 --> 0:17:40.410 still out there and then you talk about the revolution, 0:17:40.410 --> 0:17:43.680 it's still going to come in terms of artificial intelligence 0:17:44.130 --> 0:17:47.300 is that there is still a lot of uncertainty. It 0:17:47.310 --> 0:17:51.119 isn't always positive. Much of it revolves around fears around 0:17:51.119 --> 0:17:54.510 uncontrollable forces, right? In terms of what I can do. 0:17:54.930 --> 0:17:57.119 And you know, going back to your book, one of 0:17:57.119 --> 0:18:01.110 the premises is that both A.I. and climate change cause 0:18:01.560 --> 0:18:06.000 existential threats to humanity. Right? And I see this quote 0:18:06.000 --> 0:18:07.800 in front of me from your book and you ask 0:18:07.800 --> 0:18:10.080 this question and I want you to address that too. 0:18:10.560 --> 0:18:14.220 And you ask what happens if an artificial general intelligence 0:18:14.609 --> 0:18:16.800 decides that the best way to protect the Earth is 0:18:16.800 --> 0:18:20.640 to adjust the human population to sustainable numbers? And what 0:18:20.640 --> 0:18:23.970 if this number is below the current world population? Talk 0:18:23.970 --> 0:18:27.180 about that and sort of the philosophical arguments that you 0:18:27.180 --> 0:18:30.479 sort of have within yourself as to the potential and 0:18:30.480 --> 0:18:33.479 how to manage that. Those are artificial intelligence and its 0:18:33.480 --> 0:18:37.679 potential to help but also addressing the larger concerns that 0:18:37.680 --> 0:18:42.359 people may have scientist crafting this specific to climate change, 0:18:42.359 --> 0:18:42.990 for example. 0:18:43.170 --> 0:18:46.230 I personally think this is a very important question, just 0:18:46.230 --> 0:18:49.139 for the record. Okay, I want a happy, happy, happy 0:18:49.140 --> 0:18:51.810 world for everyone. So whatever we do, we have to 0:18:51.810 --> 0:18:55.949 make sure that this everybody is catered for. So the 0:18:55.950 --> 0:19:00.270 question is, what are these dangers and how can we 0:19:00.270 --> 0:19:03.240 address it? They are actually quite a bit of literature 0:19:03.900 --> 0:19:09.780 that has already done to an Austrian terms on superintelligence 0:19:09.869 --> 0:19:14.909 is a perfect example of analyzing the threats that the 0:19:14.910 --> 0:19:18.870 Superintelligence may pose. Another good book that I've read that 0:19:18.869 --> 0:19:23.250 I really love was my pick marks like 3.0 I can. 0:19:23.250 --> 0:19:27.000 We recommend this is the amazing three days of the 0:19:27.000 --> 0:19:32.430 whole artificial intelligence aspect, and the key word that is 0:19:32.430 --> 0:19:36.150 important is Google alive. So we have to make sure 0:19:36.150 --> 0:19:39.629 that the boards of technology are aligned with what we do. 0:19:40.200 --> 0:19:42.869 And this was where the problem starts, because what are 0:19:42.869 --> 0:19:46.430 our own boards? What is your Google may not be 0:19:46.560 --> 0:19:49.590 seen as mighty Google, so how do we pay for? 0:19:50.220 --> 0:19:55.109 I have decided that for our work that we start 0:19:55.109 --> 0:19:59.490 with the Sustainable Development Goals, which means basically the planet 0:19:59.490 --> 0:20:02.750 has to be sustainable. People should not be in poverty 0:20:02.750 --> 0:20:06.480 and they should have enough extreme. We should not have 0:20:06.660 --> 0:20:09.300 a lot of CO2. We should have been warned to 0:20:09.480 --> 0:20:12.510 everybody by. If you should make sure that the system 0:20:12.910 --> 0:20:16.680 tries to follow these goals, then my view is how 0:20:16.680 --> 0:20:19.440 bad can it be? Do you see the first approximation? 0:20:19.770 --> 0:20:22.169 We are sort of saying there will be an AI 0:20:22.170 --> 0:20:26.430 that starts killing people anything. I think men do that 0:20:26.430 --> 0:20:29.760 themselves enough. But we have to make sure that the 0:20:29.880 --> 0:20:34.530 right conditions for everyone, the sort of barrier that will 0:20:34.650 --> 0:20:37.620 mean that there is sort of a region, not just 0:20:37.630 --> 0:20:40.800 one point. You could imagine it would be a bit 0:20:40.800 --> 0:20:43.850 abstract that if you think about the Sustainable Development Goals 0:20:43.859 --> 0:20:47.370 at seven p.m., each of those have targets that can 0:20:47.369 --> 0:20:50.880 be quantified about and then the state of the world 0:20:50.880 --> 0:20:55.020 can be represented by one point one hundred and seventy 0:20:55.200 --> 0:21:00.180 dimensional space. Now within that space, you may have some space. 0:21:00.450 --> 0:21:03.510 The world was in that space to be good enough. 0:21:03.840 --> 0:21:08.250 But whatever happens at this point in, that space is moving. OK, 0:21:08.250 --> 0:21:11.410 so for example, if there was a massive problem, okay, 0:21:11.490 --> 0:21:13.889 we didn't have enough water for people and it would 0:21:13.890 --> 0:21:16.970 come out of it. I would say comfort zone. So 0:21:17.040 --> 0:21:20.400 you have to make sure that we constantly keep the 0:21:20.400 --> 0:21:23.669 world in this comfort zone that is found by the U.N. 0:21:23.670 --> 0:21:27.060 Development Goals. This is the way I think about too much, 0:21:27.330 --> 0:21:30.359 and maybe that was a bit too mathematical in terms 0:21:30.359 --> 0:21:32.820 of the way I described it. It is a very 0:21:32.820 --> 0:21:35.520 important question, and what we are trying to do is 0:21:35.520 --> 0:21:37.859 we not only develop the world model, we ought to 0:21:37.859 --> 0:21:41.640 develop a way to classify the work model in terms 0:21:41.640 --> 0:21:44.730 of these goal. So only then we can do forward align. 0:21:45.210 --> 0:21:48.930 The purpose of our artificial intelligence is actually to make 0:21:48.930 --> 0:21:52.950 sure that the living conditions for us are good enough 0:21:52.950 --> 0:21:54.330 to have a happy and fulfilled. 0:21:55.290 --> 0:21:58.260 I'm going to start closing off our conversation and ask 0:21:58.260 --> 0:22:01.290 really large, even larger questions, maybe to get a sense 0:22:01.290 --> 0:22:04.919 of your take on it. It seems that obviously experts 0:22:04.920 --> 0:22:07.710 in the field will say that it's is not going 0:22:07.710 --> 0:22:10.530 to be a silver bullet in dealing with climate change. 0:22:11.280 --> 0:22:14.730 But there will also be optimists, the likes of Bill Gates, 0:22:14.730 --> 0:22:18.300 for example, who believe that technology will potentially evolve to 0:22:18.300 --> 0:22:21.690 help overcome such large problems. Where do you stand in 0:22:21.690 --> 0:22:25.560 the sort of the realist versus optimist sort of spectrum? 0:22:26.010 --> 0:22:29.850 How confident are you in his potential as seen in 0:22:29.850 --> 0:22:33.050 that context, for example, in 2050, if we were to 0:22:33.060 --> 0:22:37.240 look forward next? Yes. How much of the gains in 0:22:37.240 --> 0:22:41.020 our efforts at decarbonization could actually come from gains in 0:22:41.020 --> 0:22:44.410 technology and AI? Is it something that should be part 0:22:44.410 --> 0:22:47.020 of the thinking for people and governments out there? 0:22:47.410 --> 0:22:50.630 In my opinion, absolutely. But the time period that you 0:22:50.630 --> 0:22:53.730 have mentioned is so enormous that it's very hard to 0:22:54.040 --> 0:22:58.270 think about the technological changes. I have already indicated there 0:22:58.270 --> 0:23:02.790 may be potential breakthroughs in both the efficiency, but also 0:23:02.790 --> 0:23:07.150 in the performance of these systems. And it just will 0:23:07.150 --> 0:23:10.000 help us to solve the problems. And you know, they 0:23:10.000 --> 0:23:16.030 can be used for maintenance prediction and energy optimization or 0:23:16.330 --> 0:23:19.250 basically in the classical way. We don't need to use 0:23:19.270 --> 0:23:22.570 a machine learning algorithm. You can use a classical, argumentative, 0:23:22.660 --> 0:23:25.780 similar results, maybe not as good as you can do 0:23:25.780 --> 0:23:29.050 now with the AI, but there is more than that 0:23:29.170 --> 0:23:32.560 because if you look at society at large, it's a 0:23:32.560 --> 0:23:35.530 complex system. There is a lot of information going back 0:23:35.530 --> 0:23:38.570 and forth. There is a lot of information loss you 0:23:38.590 --> 0:23:41.170 see from one people to another. If you just look 0:23:41.170 --> 0:23:44.109 at how science has been done in the past, how 0:23:44.109 --> 0:23:48.580 science is done now in students from something of the 0:23:48.580 --> 0:23:51.460 life of the ordinary person to not be, nobody else 0:23:51.460 --> 0:23:56.320 knows about it. That's over very soon. Every finding is 0:23:56.320 --> 0:24:00.550 a finding that is available to other instantaneously. Imagine the 0:24:00.550 --> 0:24:04.030 acceleration of the process of finding new things and new 0:24:04.030 --> 0:24:07.510 solutions to things, and that is all about government. If 0:24:07.510 --> 0:24:10.360 I look at how governments have to do policy, they 0:24:10.359 --> 0:24:13.300 are often in the dark and they have a report here, 0:24:13.300 --> 0:24:17.230 a report there, and they can't really base their opinions 0:24:17.230 --> 0:24:19.889 on proper fact. It's very hard for them to get 0:24:19.900 --> 0:24:23.100 it started with the smart cities and smart states and 0:24:23.109 --> 0:24:26.770 completely change. Not only they will know at any point 0:24:26.770 --> 0:24:28.750 in time, you know what the state of this, but 0:24:28.750 --> 0:24:31.990 they can induce progress in the areas we basically call 0:24:31.990 --> 0:24:36.220 these parallel worlds in your system, you can work out 0:24:36.220 --> 0:24:40.150 what if scenarios and you can implement it right away 0:24:40.480 --> 0:24:43.810 very fast. So the time for implementation time for planning 0:24:44.080 --> 0:24:47.620 all that shrinks enormously and that will help us to 0:24:47.619 --> 0:24:51.070 solve the challenges that we have. It's not just global warming, 0:24:51.070 --> 0:24:54.969 it's living in peace together. It's making sure that we 0:24:54.970 --> 0:25:00.220 build our environment, and I believe that this will help 0:25:00.220 --> 0:25:01.250 to stop same. 0:25:01.480 --> 0:25:06.520 It certainly sounds really optimistic. It sounds Huebel. But as 0:25:06.520 --> 0:25:09.190 a last question, do you therefore also in your work, 0:25:09.369 --> 0:25:13.450 have some concern that in using AI, in looking forward 0:25:13.450 --> 0:25:17.680 to the potential for A.I., humans like us may also 0:25:17.680 --> 0:25:22.840 be tempted to maybe subcontract decisions using an artificial intelligence 0:25:22.840 --> 0:25:26.260 instead of, you know, I would say, hard human choices, 0:25:26.260 --> 0:25:29.860 hard human decisions that have to be made separately from 0:25:29.859 --> 0:25:31.600 what I have to deliver. 0:25:31.630 --> 0:25:34.300 Well, I subcontract decisions in my life with software. 0:25:36.250 --> 0:25:37.770 That's always the smart thing to do. 0:25:37.780 --> 0:25:42.190 Yes, they do. But what I'm trying to say is this, 0:25:42.190 --> 0:25:45.940 of course, we will subcontract decisions to our own benefit. 0:25:46.330 --> 0:25:49.959 Every decision has consequences. You have to be aware of 0:25:49.960 --> 0:25:52.420 what they are and be willing to see them through. 0:25:52.750 --> 0:25:55.570 You know, if you look at this palest time machine 0:25:55.840 --> 0:25:59.820 and you have the Eloise and the Warlocks and the 0:25:59.830 --> 0:26:04.120 Eloise living in this fantasy world, where they don't have 0:26:04.119 --> 0:26:07.840 to care for anything and sort of slowly develop into vegetables, 0:26:07.990 --> 0:26:10.629 is that going to happen with humans? I hope not. 0:26:11.050 --> 0:26:14.199 Even if there was a super intelligent, you knew everything 0:26:14.200 --> 0:26:16.960 that can, you know, it would be not relevant to 0:26:16.960 --> 0:26:21.750 us because we still enjoy each other's company. Our thoughts, 0:26:21.940 --> 0:26:25.359 of course, would be in order. You can just look 0:26:25.359 --> 0:26:27.910 things up, but that doesn't mean that you have fully 0:26:27.910 --> 0:26:30.970 understood it and you can still think about it. Or 0:26:30.970 --> 0:26:33.840 you can think about what the system was a great agency. 0:26:35.020 --> 0:26:38.770 So I don't think it'll happen impact on our meaning 0:26:39.430 --> 0:26:43.389 as a human being. And I think this is very important. Well, 0:26:43.390 --> 0:26:46.630 Marcus Kraft, thank you very much. Thank you for being 0:26:46.630 --> 0:26:48.040 with opportunity to talk to. 0:26:50.119 --> 0:26:52.459 And thanks for listening to the climate conversation, stay up 0:26:52.460 --> 0:26:55.910 to date on CNN's coverage of climate change on CNN Asia. 0:26:55.970 --> 0:26:58.520 You can also find this and other senior podcasts on 0:26:58.520 --> 0:27:01.609 our website and on iTunes and Spotify. The team behind 0:27:01.609 --> 0:27:05.629 this podcast, Christina Robert Insulating and Erin Low. I'm Jamie 0:27:05.630 --> 0:27:06.920 Hoh again to next week.