1 00:00:02,400 --> 00:00:06,760 Speaker 1: Bloomberg Audio Studios, podcasts, radio news. 2 00:00:08,600 --> 00:00:11,080 Speaker 2: I'm Stephen Carroll and this is Here's Why, where we 3 00:00:11,119 --> 00:00:13,240 Speaker 2: take one news story and explain it in just a 4 00:00:13,280 --> 00:00:15,800 Speaker 2: few minutes with our experts here at Bloomberg. 5 00:00:21,440 --> 00:00:24,720 Speaker 1: Severe weather is making its way east after deadly storms 6 00:00:24,840 --> 00:00:25,840 Speaker 1: ravage the South. 7 00:00:26,200 --> 00:00:28,319 Speaker 2: It's a fact that we will see due to the 8 00:00:28,320 --> 00:00:33,360 Speaker 2: climate development, more events, more severe event The weather, as 9 00:00:33,400 --> 00:00:37,239 Speaker 2: always has a huge impact. Most shopping is still done 10 00:00:37,320 --> 00:00:42,159 Speaker 2: in person. I'm singlehandedly powering the gardening section of retail 11 00:00:42,280 --> 00:00:46,120 Speaker 2: sales data. Frankly, it's the center of so many conversations, 12 00:00:46,159 --> 00:00:49,599 Speaker 2: but also a key economic driver. A cold snap can 13 00:00:49,680 --> 00:00:53,000 Speaker 2: drive up energy demand. Sunny days can send shoppers out 14 00:00:53,000 --> 00:00:56,160 Speaker 2: to buy new clothes. Then there are extreme weather events 15 00:00:56,200 --> 00:01:00,120 Speaker 2: that can damage infrastructure, disrupt supply chains, and spark mass 16 00:01:00,160 --> 00:01:03,600 Speaker 2: of insurance claims. As weather gets more volatile in many 17 00:01:03,600 --> 00:01:06,360 Speaker 2: parts of the world, knowing what's on the way is 18 00:01:06,400 --> 00:01:10,559 Speaker 2: even more important. Technology has helped to improve the accuracy 19 00:01:10,600 --> 00:01:14,080 Speaker 2: of forecasts. So what does the explosion and artificial intelligence 20 00:01:14,200 --> 00:01:18,319 Speaker 2: mean for the science. Here's why AI is changing how 21 00:01:18,360 --> 00:01:23,440 Speaker 2: we predict the weather. Our weather reporter Mary Hoy joins 22 00:01:23,440 --> 00:01:25,840 Speaker 2: me now for more, Mary, before we get to the 23 00:01:25,840 --> 00:01:28,399 Speaker 2: effect of AI. Can you just bring us up to speed. 24 00:01:28,400 --> 00:01:31,160 Speaker 2: At how good have we gotten at being able to 25 00:01:31,240 --> 00:01:31,960 Speaker 2: predict the weather? 26 00:01:32,319 --> 00:01:35,959 Speaker 1: We've gotten very, very good. We've made incremental but tremendous 27 00:01:35,959 --> 00:01:39,280 Speaker 1: progress in the past decades since the very first computer 28 00:01:39,480 --> 00:01:42,440 Speaker 1: based weather forecast was made in nineteen fifty and that's 29 00:01:42,480 --> 00:01:47,560 Speaker 1: all thanks to various technological and scientific advancements of scientists 30 00:01:47,600 --> 00:01:50,720 Speaker 1: have made around the world. That's allowed us to divide 31 00:01:50,760 --> 00:01:54,560 Speaker 1: the world into smaller grids and then also divide the 32 00:01:55,000 --> 00:01:57,440 Speaker 1: air above us into more and more layers, and all 33 00:01:57,440 --> 00:01:59,240 Speaker 1: of that means that we can look at the weather 34 00:01:59,400 --> 00:02:03,400 Speaker 1: in more detail, which then increases our ability to see 35 00:02:03,400 --> 00:02:06,200 Speaker 1: further into the future. So you know, a five day 36 00:02:06,240 --> 00:02:09,120 Speaker 1: forecast now is as accurate as a three day forecast 37 00:02:09,280 --> 00:02:12,000 Speaker 1: was in two thousand. That's about a day for every 38 00:02:12,080 --> 00:02:15,200 Speaker 1: decade of technological and in scientific progress. So that's a lot 39 00:02:15,240 --> 00:02:20,000 Speaker 1: of advancement. And as the weather gets more volatile and extreme, 40 00:02:20,440 --> 00:02:23,639 Speaker 1: will be depending on this weather forecasting capabilities. 41 00:02:24,000 --> 00:02:26,840 Speaker 2: Well, I'm interested in that point. Actually, the increase in 42 00:02:26,919 --> 00:02:30,240 Speaker 2: volatile weather events seems to be something that we're observing 43 00:02:30,760 --> 00:02:33,640 Speaker 2: at pace. How does that affect weather forecasting? Does it 44 00:02:33,639 --> 00:02:34,519 Speaker 2: make it more difficult. 45 00:02:34,880 --> 00:02:39,360 Speaker 1: So weather forecasting is just fundamentally tricky. The atmosphere is complicated, 46 00:02:39,800 --> 00:02:43,840 Speaker 1: it's chaotic, it's uncertain, So every forecast has an element 47 00:02:43,840 --> 00:02:47,480 Speaker 1: of uncertainty. And that's because of incomplete observations, so not 48 00:02:47,680 --> 00:02:51,040 Speaker 1: being able to see exactly everything that's going on in 49 00:02:51,040 --> 00:02:54,720 Speaker 1: the atmosphere, and also just approximations that weather models have 50 00:02:54,800 --> 00:02:58,359 Speaker 1: to make. And now extreme weather makes that slightly trickier 51 00:02:58,560 --> 00:03:02,240 Speaker 1: and even more challenging because extreme weather, I suppose by definition, 52 00:03:02,520 --> 00:03:05,720 Speaker 1: they just have more variation variability, and they've also been 53 00:03:05,760 --> 00:03:11,000 Speaker 1: relatively rare, so that means less data and fewer observations 54 00:03:11,040 --> 00:03:13,560 Speaker 1: over time, and so we just understand these events less 55 00:03:13,600 --> 00:03:16,519 Speaker 1: well and have done less research into them. And then 56 00:03:16,800 --> 00:03:20,679 Speaker 1: these conditions can change really quickly. A hurricane, for example, 57 00:03:21,400 --> 00:03:26,120 Speaker 1: given warmer waters can rapidly intensify, and that quick change 58 00:03:26,280 --> 00:03:29,320 Speaker 1: can catch forecasters off guard. So all of that is 59 00:03:29,360 --> 00:03:32,560 Speaker 1: making it slightly trickier, even though weather forecasting itself has 60 00:03:32,600 --> 00:03:34,320 Speaker 1: never been an easy exercise. 61 00:03:34,680 --> 00:03:38,240 Speaker 2: Well, let's get to artificial intelligence. Then, how are forecasters 62 00:03:38,360 --> 00:03:41,320 Speaker 2: using AI? How can it help this process? 63 00:03:42,200 --> 00:03:45,600 Speaker 1: There are lots of steps to creating a weather forecast, 64 00:03:46,120 --> 00:03:49,240 Speaker 1: and AI can be applied to some of those steps, 65 00:03:49,360 --> 00:03:51,560 Speaker 1: or maybe all of those steps. We can take the 66 00:03:51,600 --> 00:03:55,000 Speaker 1: first step gathering realms and reams of data, because to 67 00:03:55,360 --> 00:03:57,800 Speaker 1: forecast future weather, you need to know what the present 68 00:03:57,840 --> 00:04:01,480 Speaker 1: weather's like, and that requires getting a lot of observations, 69 00:04:01,560 --> 00:04:05,880 Speaker 1: whether that's temperatures or air pressure, humidity and the like. 70 00:04:06,200 --> 00:04:10,200 Speaker 1: So AI can help by helping us gather more data 71 00:04:10,560 --> 00:04:13,520 Speaker 1: from a quantified sense, but also more types of data. 72 00:04:13,880 --> 00:04:20,440 Speaker 1: Whereas current weather models typically are restricted to meteorological observations, 73 00:04:20,480 --> 00:04:23,520 Speaker 1: AI models can now allow us to expand it to 74 00:04:23,640 --> 00:04:28,240 Speaker 1: say maintenance logs for power grids, or even news articles, 75 00:04:28,560 --> 00:04:32,680 Speaker 1: just really gathering more and more information to inform models 76 00:04:32,680 --> 00:04:37,200 Speaker 1: with to then create perhaps hopefully more accurate forecasts. 77 00:04:37,800 --> 00:04:41,680 Speaker 2: How widespread is the use of AI and forecasting currently. 78 00:04:42,080 --> 00:04:47,240 Speaker 1: So a key intergovernmental weather forecasting agency called the European 79 00:04:47,279 --> 00:04:51,960 Speaker 1: Center for Medium Range Weather Forecasting or ECMWF. They've taken 80 00:04:52,240 --> 00:04:54,960 Speaker 1: their AI model into operations earlier this year and they're 81 00:04:55,000 --> 00:04:59,200 Speaker 1: running it side by side with their existing traditional physics 82 00:04:59,200 --> 00:05:03,240 Speaker 1: based the model, and they're soon to launch another iteration 83 00:05:03,320 --> 00:05:05,960 Speaker 1: of that AI model into operations too. So there's lots 84 00:05:05,960 --> 00:05:09,360 Speaker 1: going on in this world. So the ECMWF. They're also 85 00:05:09,440 --> 00:05:14,400 Speaker 1: running AI models from different providers, including China's Huawei the 86 00:05:14,480 --> 00:05:19,480 Speaker 1: tech giant, Google's graph Cast model, Microsoft's Aurora model, and 87 00:05:19,680 --> 00:05:22,599 Speaker 1: China's own Weather Bureau itself is also testing about a 88 00:05:22,640 --> 00:05:26,479 Speaker 1: dozen AI weather models, so a lot of experimentation going on. 89 00:05:26,560 --> 00:05:31,160 Speaker 1: There were probably going to see AI models and traditional 90 00:05:31,200 --> 00:05:34,200 Speaker 1: weather models working in tandem in the years to come, 91 00:05:34,320 --> 00:05:38,760 Speaker 1: rather than seeing AI can completely replace these traditional numerical 92 00:05:39,040 --> 00:05:40,760 Speaker 1: weather model forecasting methods. 93 00:05:41,240 --> 00:05:43,960 Speaker 2: Yeah, I'm curious if there are risks to using AI 94 00:05:44,120 --> 00:05:47,280 Speaker 2: for weather forecasting. Is there things that technology might miss 95 00:05:47,760 --> 00:05:49,920 Speaker 2: if we see a greater implementation of it. 96 00:05:50,960 --> 00:05:53,880 Speaker 1: The way a lot of these AI models are trained 97 00:05:54,000 --> 00:05:59,560 Speaker 1: right now is on historical climate data from decades and 98 00:05:59,600 --> 00:06:03,039 Speaker 1: decades observations in analyzes. But the thing is that the 99 00:06:03,080 --> 00:06:05,680 Speaker 1: past is never perfect predictor of the future, and so 100 00:06:06,279 --> 00:06:09,960 Speaker 1: if we're about to see in a warming climate, more 101 00:06:10,080 --> 00:06:15,120 Speaker 1: extreme weather that previously is not reflected in these historical 102 00:06:15,200 --> 00:06:17,800 Speaker 1: data sets, then there is a risk that AI weather 103 00:06:17,880 --> 00:06:21,760 Speaker 1: models will say underestimates maybe the intensity of a typhoon 104 00:06:21,839 --> 00:06:25,960 Speaker 1: or a tropical cyclone, even as it improves the forecasts 105 00:06:26,000 --> 00:06:28,520 Speaker 1: of a storm's track, for example. So there are some 106 00:06:28,640 --> 00:06:31,720 Speaker 1: risks there, and it's definitely an open and ongoing feel 107 00:06:31,720 --> 00:06:32,320 Speaker 1: of the research. 108 00:06:32,800 --> 00:06:35,240 Speaker 2: So what's the next big development we should be watching 109 00:06:35,279 --> 00:06:36,360 Speaker 2: out for in this area? 110 00:06:37,360 --> 00:06:40,760 Speaker 1: This is less one single big development but rather trend 111 00:06:40,760 --> 00:06:43,760 Speaker 1: and I'd be looking for kind of a shifting distribution 112 00:06:43,960 --> 00:06:47,520 Speaker 1: right of roles in this whole global weather enterprise. Are 113 00:06:47,520 --> 00:06:52,360 Speaker 1: public weather agencies which we've long depended on for weather forecasts, 114 00:06:52,640 --> 00:06:55,560 Speaker 1: are they going to play a slightly different role now 115 00:06:56,000 --> 00:07:00,640 Speaker 1: that private companies, tech firms and also small players increasingly 116 00:07:00,720 --> 00:07:04,640 Speaker 1: jumping into this world of creating and providing weather forecasts 117 00:07:04,760 --> 00:07:07,159 Speaker 1: at a more niche level. So it'll be interesting to 118 00:07:07,200 --> 00:07:10,600 Speaker 1: see how that division of labor between the public forecasters 119 00:07:10,600 --> 00:07:12,280 Speaker 1: and more private players will play out. 120 00:07:12,960 --> 00:07:15,200 Speaker 2: Okay, Mary Hoy, our weather reporter. Thank you very much 121 00:07:15,240 --> 00:07:18,640 Speaker 2: for joining us. For more explanations like this from our 122 00:07:18,680 --> 00:07:21,679 Speaker 2: team of three thousand journalists and analysts around the world, 123 00:07:21,800 --> 00:07:25,960 Speaker 2: go to Bloomberg dot com slash explainers. I'm Stephen Carroll. 124 00:07:26,080 --> 00:07:28,520 Speaker 2: This is here's why. I'll be back next week with more. 125 00:07:28,720 --> 00:07:29,520 Speaker 2: Thanks for listening.