WEBVTT 1 00:00:00.000 --> 00:00:00.900 2 00:00:00.900 --> 00:00:03.380 SAM RANSBOTHAM: Many tasks could benefit 3 00:00:03.380 --> 00:00:06.280 from advanced technologies, but how can organizations 4 00:00:06.280 --> 00:00:09.320 use emerging technologies in high-stakes situations? 5 00:00:09.320 --> 00:00:11.410 Find out how the aerospace industry 6 00:00:11.410 --> 00:00:14.020 is using AI in today's episode. 7 00:00:14.020 --> 00:00:16.230 HELEN LEE: I'm Helen Lee from Boeing, 8 00:00:16.230 --> 00:00:19.000 and you're listening to Me, Myself, and AI. 9 00:00:19.000 --> 00:00:21.910 SAM RANSBOTHAM: Welcome to Me, Myself, and AI, 10 00:00:21.910 --> 00:00:24.930 a podcast on artificial intelligence in business. 11 00:00:24.930 --> 00:00:28.690 Each episode, we introduce you to someone innovating with AI. 12 00:00:28.690 --> 00:00:32.980 I'm Sam Ransbotham, professor of analytics at Boston College. 13 00:00:32.980 --> 00:00:36.520 I'm also the AI and business strategy guest editor 14 00:00:36.520 --> 00:00:38.020 at MIT Sloan Management Review. 15 00:00:38.020 --> 00:00:38.298 16 00:00:38.298 --> 00:00:40.340 SHERVIN KHODABANDEH: And I'm Shervin Khodabandeh, 17 00:00:40.340 --> 00:00:44.410 senior partner with BCG, and I colead BCG's AI practice 18 00:00:44.410 --> 00:00:45.400 in North America. 19 00:00:45.400 --> 00:00:49.910 Together, MIT SMR and BCG have been researching and publishing 20 00:00:49.910 --> 00:00:52.840 on AI for six years, interviewing hundreds 21 00:00:52.840 --> 00:00:54.960 of practitioners and surveying thousands 22 00:00:54.960 --> 00:00:58.440 of companies on what it takes to build and to deploy and scale 23 00:00:58.440 --> 00:01:00.910 AI capabilities and really transform 24 00:01:00.910 --> 00:01:02.230 the way organizations operate. 25 00:01:02.230 --> 00:01:04.950 SAM RANSBOTHAM: Shervin and I are 26 00:01:04.950 --> 00:01:08.490 excited to be talking today with Helen Lee, regional director 27 00:01:08.490 --> 00:01:10.780 of air traffic management and airport programs 28 00:01:10.780 --> 00:01:12.777 in China for The Boeing Company. 29 00:01:12.777 --> 00:01:14.860 Helen, thanks for taking the time to talk with us. 30 00:01:14.860 --> 00:01:15.360 Welcome. 31 00:01:15.360 --> 00:01:16.818 HELEN LEE: Thank you for having me. 32 00:01:16.818 --> 00:01:18.570 SAM RANSBOTHAM: Let's get started. 33 00:01:18.570 --> 00:01:21.182 Helen, can you tell us about your current role at Boeing? 34 00:01:21.182 --> 00:01:22.420 35 00:01:22.420 --> 00:01:25.390 HELEN LEE: I'm currently working at Boeing China in the Beijing 36 00:01:25.390 --> 00:01:26.360 office. 37 00:01:26.360 --> 00:01:29.550 My main responsibility is to oversee 38 00:01:29.550 --> 00:01:32.500 Boeing's airport and air traffic management 39 00:01:32.500 --> 00:01:33.760 [ATM] programs in China. 40 00:01:33.760 --> 00:01:38.900 What we try to do here is to improve airport and airspace 41 00:01:38.900 --> 00:01:42.210 operational efficiency and at the same time 42 00:01:42.210 --> 00:01:44.390 enhance flight safety. 43 00:01:44.390 --> 00:01:49.480 I've been doing consulting for airports and ATM for 10 years, 44 00:01:49.480 --> 00:01:53.400 and the interesting part is every project 45 00:01:53.400 --> 00:01:56.240 is very different; I'm able to work with different groups 46 00:01:56.240 --> 00:01:57.230 of people. 47 00:01:57.230 --> 00:02:01.050 Air traffic management is not the core business for Boeing, 48 00:02:01.050 --> 00:02:04.890 but what we really want is to help our customers. 49 00:02:04.890 --> 00:02:07.510 And because airlines are our customers, 50 00:02:07.510 --> 00:02:10.570 we want them operate more efficiently 51 00:02:10.570 --> 00:02:14.310 and also enhance their flight safety. 52 00:02:14.310 --> 00:02:17.790 Other than the design and manufacture of aircraft, 53 00:02:17.790 --> 00:02:20.630 we also provide a lot of services 54 00:02:20.630 --> 00:02:22.720 to our airline customers. 55 00:02:22.720 --> 00:02:25.700 It's kind of like the entire ecosystem 56 00:02:25.700 --> 00:02:29.020 that Boeing and actually other partners 57 00:02:29.020 --> 00:02:30.660 are working on together. 58 00:02:30.660 --> 00:02:35.620 For example, we provide flight planning to airline customers, 59 00:02:35.620 --> 00:02:40.010 and that will help them to better plan their flights 60 00:02:40.010 --> 00:02:44.180 and optimize their flight paths, and that also 61 00:02:44.180 --> 00:02:47.050 will reduce their fuel consumption 62 00:02:47.050 --> 00:02:49.420 and emission of carbon dioxide. 63 00:02:49.420 --> 00:02:52.120 SHERVIN KHODABANDEH: Air traffic management 64 00:02:52.120 --> 00:02:54.530 has always been quite fascinating to me 65 00:02:54.530 --> 00:02:57.460 because I always think, there's tens of thousands 66 00:02:57.460 --> 00:03:00.270 of aircraft at any given time in the air, 67 00:03:00.270 --> 00:03:02.950 and you have different systems trying 68 00:03:02.950 --> 00:03:06.200 to do air traffic management or air traffic 69 00:03:06.200 --> 00:03:09.300 control for that aircraft. 70 00:03:09.300 --> 00:03:12.840 It feels like it's very data centric. 71 00:03:12.840 --> 00:03:15.390 It's also somewhat, maybe, chaotic, 72 00:03:15.390 --> 00:03:17.810 where unpredictable things can happen. 73 00:03:17.810 --> 00:03:19.620 And then you probably have -- 74 00:03:19.620 --> 00:03:22.800 I don't know, you tell us -- some interaction effects 75 00:03:22.800 --> 00:03:27.130 between different ATMs, or air traffic management systems, 76 00:03:27.130 --> 00:03:32.210 but can you just educate us and our audience a bit on what it 77 00:03:32.210 --> 00:03:35.600 is and how it works and how sophisticated and complicated 78 00:03:35.600 --> 00:03:36.100 it is? 79 00:03:36.100 --> 00:03:40.200 HELEN LEE: So you probably all know that a flight usually 80 00:03:40.200 --> 00:03:43.430 takes six phases: testing on the ground 81 00:03:43.430 --> 00:03:47.240 and either testing in or testing out; and then takeoff 82 00:03:47.240 --> 00:03:51.720 from the runway; and then climbing; and then en route; 83 00:03:51.720 --> 00:03:55.210 and then start descent; and landing. 84 00:03:55.210 --> 00:03:57.370 So there are six phases of it. 85 00:03:57.370 --> 00:04:00.630 And in each phase, there are air traffic controllers 86 00:04:00.630 --> 00:04:02.550 controlling the aircraft. 87 00:04:02.550 --> 00:04:05.050 General aviation in the U.S. is very different. 88 00:04:05.050 --> 00:04:08.290 [A plane] could fly in controlled airspace, 89 00:04:08.290 --> 00:04:10.710 but for most of our commercial flights, 90 00:04:10.710 --> 00:04:13.830 you need to go through the six phases. 91 00:04:13.830 --> 00:04:16.910 First, you have a controller at the ground, 92 00:04:16.910 --> 00:04:20.089 and you probably see the airport's high tower -- 93 00:04:20.089 --> 00:04:21.690 that's the ATC tower. 94 00:04:21.690 --> 00:04:25.130 They have the ground controller that controls the ground 95 00:04:25.130 --> 00:04:28.490 movement, and then there's the tower controller, who controls 96 00:04:28.490 --> 00:04:30.250 the takeoff and landing. 97 00:04:30.250 --> 00:04:35.020 And once the aircraft climbs to a certain altitude, 98 00:04:35.020 --> 00:04:38.060 you will be handed over to an approach controller, 99 00:04:38.060 --> 00:04:40.910 or departure controller; you can say that. 100 00:04:40.910 --> 00:04:43.790 So that controller will handle the aircraft 101 00:04:43.790 --> 00:04:48.490 to about 30 to 60 nautical miles from the airport, 102 00:04:48.490 --> 00:04:51.380 and then you are passed to the en route controller. 103 00:04:51.380 --> 00:04:54.060 And then the en route controller usually 104 00:04:54.060 --> 00:04:58.380 has to go through several centers, and in each center, 105 00:04:58.380 --> 00:05:03.780 there could be many sectors, and each sector will usually 106 00:05:03.780 --> 00:05:05.940 be managed by one controller. 107 00:05:05.940 --> 00:05:09.710 And so they hand off one by one, all the way 108 00:05:09.710 --> 00:05:11.720 through to the landing site. 109 00:05:11.720 --> 00:05:14.190 So that's how that works. 110 00:05:14.190 --> 00:05:17.580 But I would have to say, AI is still not 111 00:05:17.580 --> 00:05:20.420 widely used in ATM systems. 112 00:05:20.420 --> 00:05:24.310 There are many, many challenges; we can talk about that later. 113 00:05:24.310 --> 00:05:29.150 They're still using voice communication mostly. 114 00:05:29.150 --> 00:05:32.160 And so that's the part that hasn't 115 00:05:32.160 --> 00:05:36.540 been replaced by this data-link communication yet. 116 00:05:36.540 --> 00:05:39.470 But we're moving in that direction. 117 00:05:39.470 --> 00:05:44.150 One application that's widely being used 118 00:05:44.150 --> 00:05:47.970 is computer vision, like image recognition. 119 00:05:47.970 --> 00:05:52.380 And so, for example, what we are doing right now is, 120 00:05:52.380 --> 00:05:58.240 in one of the studies, we're [using] that to recognize wake 121 00:05:58.240 --> 00:06:03.480 turbulence -- the wake vortex right after a landing aircraft, 122 00:06:03.480 --> 00:06:05.540 or maybe for an aircraft in approach. 123 00:06:05.540 --> 00:06:11.360 And so we are using a lidar machine to observe that wake, 124 00:06:11.360 --> 00:06:16.780 and then we use the AI algorithm to help us to capture those. 125 00:06:16.780 --> 00:06:21.260 And so we can train the machine to recognize 126 00:06:21.260 --> 00:06:25.130 the location and the strength of that wake vortex. 127 00:06:25.130 --> 00:06:28.830 That would be something we might be applying in the future 128 00:06:28.830 --> 00:06:32.330 to shorten the wake turbulence separation. 129 00:06:32.330 --> 00:06:35.840 Another one is, as I mentioned earlier, 130 00:06:35.840 --> 00:06:37.950 about speech recognition. 131 00:06:37.950 --> 00:06:42.310 That is something that we are doing a lot of research on -- 132 00:06:42.310 --> 00:06:46.320 and not just Boeing, but other parts of the industry -- 133 00:06:46.320 --> 00:06:49.590 to speech-recognize the cockpit conversation with 134 00:06:49.590 --> 00:06:50.710 the controller. 135 00:06:50.710 --> 00:06:54.560 Because some of the instructions from the controller 136 00:06:54.560 --> 00:06:59.800 are kind of the same from one aircraft to another aircraft, 137 00:06:59.800 --> 00:07:03.730 that part may be able to be digitized and just 138 00:07:03.730 --> 00:07:07.330 use a data form and have a display in the cockpit instead 139 00:07:07.330 --> 00:07:11.330 of having the controller repeat it all the time. 140 00:07:11.330 --> 00:07:16.140 And if an aircraft has to be rerouted to another path, 141 00:07:16.140 --> 00:07:19.120 that would be another thing that can be digitized 142 00:07:19.120 --> 00:07:21.260 using speech recognition. 143 00:07:21.260 --> 00:07:25.160 The benefit of it is, if it's all digitized, 144 00:07:25.160 --> 00:07:27.400 it's all coming from the controller, 145 00:07:27.400 --> 00:07:30.130 and then we don't need the pilot to punch 146 00:07:30.130 --> 00:07:33.900 in the paths, the waypoints, into the system 147 00:07:33.900 --> 00:07:37.480 and they can be uploaded directly to the flight 148 00:07:37.480 --> 00:07:38.690 management system. 149 00:07:38.690 --> 00:07:41.900 So they will avoid some of the errors made by the pilots. 150 00:07:41.900 --> 00:07:42.400 151 00:07:42.400 --> 00:07:44.440 SHERVIN KHODABANDEH: So we're hearing 152 00:07:44.440 --> 00:07:46.680 there are a lot of potential applications 153 00:07:46.680 --> 00:07:49.910 but not yet --widespread use. 154 00:07:49.910 --> 00:07:51.900 What are some of the reasons for that? 155 00:07:51.900 --> 00:07:54.585 You said voice recognition, for example, right? 156 00:07:54.585 --> 00:07:55.210 HELEN LEE: Yes. 157 00:07:55.210 --> 00:07:57.610 SHERVIN KHODABANDEH: Like my Siri 158 00:07:57.610 --> 00:07:59.750 doesn't have any problems recognizing my voice, 159 00:07:59.750 --> 00:08:02.320 although I have heard, on YouTube, some 160 00:08:02.320 --> 00:08:05.710 of the conversations between air traffic control and pilots, 161 00:08:05.710 --> 00:08:08.800 and it's as strange to me as deciphering 162 00:08:08.800 --> 00:08:11.000 my doctor's prescription. 163 00:08:11.000 --> 00:08:12.802 But tell us more: Why is that hard? 164 00:08:12.802 --> 00:08:13.980 165 00:08:13.980 --> 00:08:16.340 HELEN LEE: One is the reliability. 166 00:08:16.340 --> 00:08:18.460 You know, how reliable [are they] 167 00:08:18.460 --> 00:08:20.120 when they recognize the voice? 168 00:08:20.120 --> 00:08:23.670 Because it cannot make any mistakes. 169 00:08:23.670 --> 00:08:25.880 In the air traffic management world, 170 00:08:25.880 --> 00:08:28.210 there's zero tolerance for mistakes. 171 00:08:28.210 --> 00:08:31.160 So it's not like your Siri, where 172 00:08:31.160 --> 00:08:34.500 if you make a mistake, that's fine; you just say it again. 173 00:08:34.500 --> 00:08:37.210 But not here; you cannot do it. 174 00:08:37.210 --> 00:08:38.960 And so that's one thing. 175 00:08:38.960 --> 00:08:40.750 And the other one that I was saying, 176 00:08:40.750 --> 00:08:42.730 probably with data link communication, 177 00:08:42.730 --> 00:08:46.910 is the security -- whether we have a very secure environment 178 00:08:46.910 --> 00:08:51.160 that nobody could hack into or things like that. 179 00:08:51.160 --> 00:08:54.890 That's why so far, we haven't seen any application being 180 00:08:54.890 --> 00:08:55.930 certified yet. 181 00:08:55.930 --> 00:08:59.070 SHERVIN KHODABANDEH: If I'm understanding you correctly, 182 00:08:59.070 --> 00:09:04.740 one of the biggest hurdles is the need for absolute 183 00:09:04.740 --> 00:09:09.550 precision, zero error tolerance, and just how much is at stake 184 00:09:09.550 --> 00:09:10.150 that it's -- 185 00:09:10.150 --> 00:09:10.560 HELEN LEE: Correct. 186 00:09:10.560 --> 00:09:12.988 SHERVIN KHODABANDEH: -- maybe unlike many other things. 187 00:09:12.988 --> 00:09:14.780 SAM RANSBOTHAM: Yes, Shervin, I think a lot 188 00:09:14.780 --> 00:09:17.550 that we talk about is corporate application, where 189 00:09:17.550 --> 00:09:20.260 people make a recommendation or a loan approval. 190 00:09:20.260 --> 00:09:23.840 It's not real time; it's not critical in the moment, 191 00:09:23.840 --> 00:09:26.790 whereas this is a very different scenario. 192 00:09:26.790 --> 00:09:29.067 But not all the scenarios are very different. 193 00:09:29.067 --> 00:09:30.650 Like, for example, the wake turbulence 194 00:09:30.650 --> 00:09:33.200 is something that doesn't have to be real time. 195 00:09:33.200 --> 00:09:36.370 That could be an after-the-fact analysis; 196 00:09:36.370 --> 00:09:38.465 runway configuration could be after the fact. 197 00:09:38.465 --> 00:09:40.840 SHERVIN KHODABANDEH: And it's not just real time, either. 198 00:09:40.840 --> 00:09:44.510 It's like, if you think about algorithmic trading that's 199 00:09:44.510 --> 00:09:46.990 going on, it's near real time, or the credit card 200 00:09:46.990 --> 00:09:48.430 authorization is real time. 201 00:09:48.430 --> 00:09:53.290 I think it's real time, but also, how much is at stake -- 202 00:09:53.290 --> 00:09:54.980 like, [what is] the cost of being wrong? 203 00:09:54.980 --> 00:09:58.280 Like with image recognition, for example, or video, 204 00:09:58.280 --> 00:10:01.960 like in medical applications, you still have a doctor, 205 00:10:01.960 --> 00:10:07.540 and if there's a mistake, God forbid, it's one life; 206 00:10:07.540 --> 00:10:09.890 it's not hundreds of lives. 207 00:10:09.890 --> 00:10:12.890 Are we getting it, that it's really 208 00:10:12.890 --> 00:10:14.780 the gravity of the situation? 209 00:10:14.780 --> 00:10:16.700 Is that what prevents these things 210 00:10:16.700 --> 00:10:22.420 from being widely adopted in the flight life cycle? 211 00:10:22.420 --> 00:10:25.570 HELEN LEE: Actually, in the last couple decades, 212 00:10:25.570 --> 00:10:30.980 the industry has been really preparing for this technology 213 00:10:30.980 --> 00:10:34.370 to be applied in this industry, so a lot of work 214 00:10:34.370 --> 00:10:39.770 has been done to [support] the automation of the system. 215 00:10:39.770 --> 00:10:41.250 So that's one part. 216 00:10:41.250 --> 00:10:44.040 Now we know most of the aircraft, especially 217 00:10:44.040 --> 00:10:48.360 new aircraft the advantage to using this in China 218 00:10:48.360 --> 00:10:52.280 is, they have almost all new aircraft in China, 219 00:10:52.280 --> 00:10:55.410 because most of their aircraft is less than 10 years old. 220 00:10:55.410 --> 00:10:58.140 So that means they're all equipped with the latest 221 00:10:58.140 --> 00:10:59.570 technology onboard. 222 00:10:59.570 --> 00:11:01.880 And so that's one thing. 223 00:11:01.880 --> 00:11:05.350 And also, their control surfaces -- 224 00:11:05.350 --> 00:11:08.590 they use the most advanced technology as well. 225 00:11:08.590 --> 00:11:12.380 I would say the basics for AI to be 226 00:11:12.380 --> 00:11:18.620 applied is to have some of the automation system be there. 227 00:11:18.620 --> 00:11:20.770 One thing that we are also doing you 228 00:11:20.770 --> 00:11:22.940 know, aircraft can do [autopilot]. 229 00:11:22.940 --> 00:11:26.840 Before, many times we used the procedure 230 00:11:26.840 --> 00:11:30.410 of landing and departing from an airport. 231 00:11:30.410 --> 00:11:35.570 But now, we are really promoting performance-based navigation. 232 00:11:35.570 --> 00:11:37.400 And since most of aircraft already 233 00:11:37.400 --> 00:11:40.230 have that navigation system equipped, 234 00:11:40.230 --> 00:11:45.700 we are able to give an aircraft a more precise route for it 235 00:11:45.700 --> 00:11:48.780 to climb out or descend to an airport. 236 00:11:48.780 --> 00:11:52.600 And that means it'll be much easier later on if we 237 00:11:52.600 --> 00:11:55.310 try to manage those aircraft. 238 00:11:55.310 --> 00:11:59.280 So that's one thing: Kind of create a base that so we 239 00:11:59.280 --> 00:12:04.070 can build upon it and use more advanced technology 240 00:12:04.070 --> 00:12:05.440 in this system. 241 00:12:05.440 --> 00:12:08.990 And there are many new studies coming out, 242 00:12:08.990 --> 00:12:10.860 and there are road maps and plans 243 00:12:10.860 --> 00:12:14.270 for using AI in air traffic management. 244 00:12:14.270 --> 00:12:17.380 So I would say in the next decade, 245 00:12:17.380 --> 00:12:22.610 we'll probably see a lot more things come up using AI. 246 00:12:22.610 --> 00:12:24.950 I would have to say another challenge we 247 00:12:24.950 --> 00:12:29.210 have in using AI is, usually the AI, 248 00:12:29.210 --> 00:12:32.630 if you apply it to an air traffic management system, 249 00:12:32.630 --> 00:12:37.190 we might rely on a knowledge-based expert system. 250 00:12:37.190 --> 00:12:41.660 So it is very hard to build a good expert system; 251 00:12:41.660 --> 00:12:44.900 especially in different environments, 252 00:12:44.900 --> 00:12:48.730 their expert system may be completely different 253 00:12:48.730 --> 00:12:50.400 because their operation is different. 254 00:12:50.400 --> 00:12:51.900 They may have a [different] terrain, 255 00:12:51.900 --> 00:12:54.750 they may have different runway configurations and all that. 256 00:12:54.750 --> 00:12:58.960 So that's another part: You cannot just build one expert 257 00:12:58.960 --> 00:13:01.920 system to use everywhere. 258 00:13:01.920 --> 00:13:04.720 SAM RANSBOTHAM: So there's a lot of preparation work, 259 00:13:04.720 --> 00:13:06.210 and what struck me as particularly 260 00:13:06.210 --> 00:13:08.040 interesting about what you are saying 261 00:13:08.040 --> 00:13:11.390 is how coordinated that needs to be with lots 262 00:13:11.390 --> 00:13:14.480 of different people, lots of different organizations, 263 00:13:14.480 --> 00:13:15.760 different airlines. 264 00:13:15.760 --> 00:13:20.540 This isn't just a thing that one organization can put in place 265 00:13:20.540 --> 00:13:22.810 and dictate to their people that they use. 266 00:13:22.810 --> 00:13:25.655 It's something that has to coordinate across lots 267 00:13:25.655 --> 00:13:26.780 of different organizations. 268 00:13:26.780 --> 00:13:30.100 With equipment like airplanes, you can't just [say], 269 00:13:30.100 --> 00:13:32.760 "Oh, well, let's just all get new airplanes next week so they 270 00:13:32.760 --> 00:13:34.410 have the new technology." 271 00:13:34.410 --> 00:13:35.920 So it's complicated. 272 00:13:35.920 --> 00:13:36.143 273 00:13:36.143 --> 00:13:37.810 HELEN LEE: You're very right about that. 274 00:13:37.810 --> 00:13:39.960 And that's why everybody's working 275 00:13:39.960 --> 00:13:43.550 on what we call SWIM, that's "systemwide information 276 00:13:43.550 --> 00:13:44.480 management." 277 00:13:44.480 --> 00:13:48.610 So that means we are able to share information 278 00:13:48.610 --> 00:13:52.570 between different players; that could be the airline, 279 00:13:52.570 --> 00:13:54.780 could be pilots in the cockpit, could 280 00:13:54.780 --> 00:13:56.540 be air traffic controllers. 281 00:13:56.540 --> 00:13:59.050 And then we will have the meteorology data, 282 00:13:59.050 --> 00:14:01.320 and everything will come together. 283 00:14:01.320 --> 00:14:05.100 So everything will be shared within the system. 284 00:14:05.100 --> 00:14:08.680 Different players will able to see the information 285 00:14:08.680 --> 00:14:12.060 and data they need to better operate their own system. 286 00:14:12.060 --> 00:14:15.710 SAM RANSBOTHAM: You pointed to a lot of forward-looking aspects 287 00:14:15.710 --> 00:14:17.380 of artificial intelligence. 288 00:14:17.380 --> 00:14:19.570 Is there something you're using right now? 289 00:14:19.570 --> 00:14:22.830 Is there something that Boeing is doing right now that maybe 290 00:14:22.830 --> 00:14:26.210 we don't know about or that is behind the scenes that's 291 00:14:26.210 --> 00:14:27.510 hard for people to see? 292 00:14:27.510 --> 00:14:29.360 What kinds of artificial intelligence 293 00:14:29.360 --> 00:14:31.050 are currently in use right now? 294 00:14:31.050 --> 00:14:34.030 HELEN LEE: One thing that's coming 295 00:14:34.030 --> 00:14:39.490 very close to application is using image recognition. 296 00:14:39.490 --> 00:14:43.330 I listened to one of your peer's [episodes] -- 297 00:14:43.330 --> 00:14:45.450 [Gina Chung] from DHL. 298 00:14:45.450 --> 00:14:49.030 She mentioned a similar technology. 299 00:14:49.030 --> 00:14:51.880 So, for example, when an aircraft is coming in, we can 300 00:14:51.880 --> 00:14:56.110 use a robot camera and to take pictures of the aircraft, 301 00:14:56.110 --> 00:14:57.830 to take pictures of the fuselage, 302 00:14:57.830 --> 00:15:02.580 to see if there's any damage and then use AI to recognize 303 00:15:02.580 --> 00:15:05.860 whether that's an important thing that we need to take care 304 00:15:05.860 --> 00:15:09.195 of -- whether it needs to go into the hangar to be fixed, 305 00:15:09.195 --> 00:15:10.320 you know, things like that. 306 00:15:10.320 --> 00:15:12.688 SAM RANSBOTHAM: That seems like a great application. 307 00:15:12.688 --> 00:15:14.230 HELEN LEE: Yes, but before, you know, 308 00:15:14.230 --> 00:15:18.350 you have to have a human being walk around the aircraft 309 00:15:18.350 --> 00:15:21.685 to identify all those and then make decisions. 310 00:15:21.685 --> 00:15:23.810 SAM RANSBOTHAM: I think what's difficult about some 311 00:15:23.810 --> 00:15:26.187 of those things is what you don't notice 312 00:15:26.187 --> 00:15:27.270 is when they don't happen. 313 00:15:27.270 --> 00:15:30.670 Let's say you do a great job of inspecting the plane beforehand 314 00:15:30.670 --> 00:15:32.970 and finding a problem and preventing it, 315 00:15:32.970 --> 00:15:36.070 or recognizing a part needs service before people 316 00:15:36.070 --> 00:15:37.570 are actually on the plane. 317 00:15:37.570 --> 00:15:40.620 These are not things that people notice. 318 00:15:40.620 --> 00:15:42.420 You only notice when it doesn't work. 319 00:15:42.420 --> 00:15:42.683 320 00:15:42.683 --> 00:15:43.475 HELEN LEE: Correct! 321 00:15:43.475 --> 00:15:43.582 322 00:15:43.582 --> 00:15:45.790 SAM RANSBOTHAM: It's the classic engineering problem. 323 00:15:45.790 --> 00:15:47.710 HELEN LEE: You're absolutely right. 324 00:15:47.710 --> 00:15:50.130 We are in the process of collecting data 325 00:15:50.130 --> 00:15:53.390 because we need a lot of data to train the machine. 326 00:15:53.390 --> 00:15:57.170 And the most important part is to collect all those data. 327 00:15:57.170 --> 00:16:00.980 And nowadays, with the new aircraft, like the 787, 328 00:16:00.980 --> 00:16:03.980 there's a lot of data we can collect. 329 00:16:03.980 --> 00:16:09.250 It's not like the older 747 that was built decades ago. 330 00:16:09.250 --> 00:16:12.390 But the new aircraft that we make today, 331 00:16:12.390 --> 00:16:14.420 we are able to have a lot of data, 332 00:16:14.420 --> 00:16:17.840 and then those data will help us to analyze 333 00:16:17.840 --> 00:16:19.270 the health of the aircraft. 334 00:16:19.270 --> 00:16:21.020 SAM RANSBOTHAM: That seems great. 335 00:16:21.020 --> 00:16:22.880 So, Helen, we have a new segment, 336 00:16:22.880 --> 00:16:26.080 and we ask our guests a series of rapid-fire questions. 337 00:16:26.080 --> 00:16:27.900 So just answer the first response 338 00:16:27.900 --> 00:16:28.940 that comes to your mind. 339 00:16:28.940 --> 00:16:30.690 You don't have to think about it too much. 340 00:16:30.690 --> 00:16:32.520 Just [go with] your first reaction. 341 00:16:32.520 --> 00:16:34.880 So, what's been your proudest moment of using 342 00:16:34.880 --> 00:16:36.460 artificial intelligence? 343 00:16:36.460 --> 00:16:37.465 344 00:16:37.465 --> 00:16:39.840 HELEN LEE: It's hard, because I don't use that every day. 345 00:16:39.840 --> 00:16:40.150 [Laughs.] 346 00:16:40.150 --> 00:16:42.010 SHERVIN KHODABANDEH: I think that was the response. 347 00:16:42.010 --> 00:16:42.410 [Laughs.] 348 00:16:42.410 --> 00:16:43.410 SAM RANSBOTHAM: Exactly. 349 00:16:43.410 --> 00:16:44.470 That may be your answer. 350 00:16:44.470 --> 00:16:44.874 351 00:16:44.874 --> 00:16:46.916 SHERVIN KHODABANDEH: No response is the response. 352 00:16:46.916 --> 00:16:47.493 353 00:16:47.493 --> 00:16:48.160 HELEN LEE: Yeah. 354 00:16:48.160 --> 00:16:48.700 SAM RANSBOTHAM: OK. 355 00:16:48.700 --> 00:16:51.110 Well, what worries you about artificial intelligence? 356 00:16:51.110 --> 00:16:52.550 357 00:16:52.550 --> 00:16:54.570 HELEN LEE: The challenge we were talking about. 358 00:16:54.570 --> 00:16:59.800 You know, how safe it can be if it really is applied to the air 359 00:16:59.800 --> 00:17:01.424 traffic control environment. 360 00:17:01.424 --> 00:17:01.717 361 00:17:01.717 --> 00:17:03.800 SAM RANSBOTHAM: What's your favorite activity that 362 00:17:03.800 --> 00:17:05.790 involves no technology? 363 00:17:05.790 --> 00:17:08.290 HELEN LEE: Oh, that's something I couldn't do now: 364 00:17:08.290 --> 00:17:12.422 I used to do kayaking when I was living in Atlanta. 365 00:17:12.422 --> 00:17:13.380 SAM RANSBOTHAM: Oh, OK. 366 00:17:13.380 --> 00:17:14.910 Well, I'm actually from Atlanta, so yeah. 367 00:17:14.910 --> 00:17:15.493 HELEN LEE: Oh! 368 00:17:15.493 --> 00:17:17.950 SAM RANSBOTHAM: We probably kayaked the same waters then. 369 00:17:17.950 --> 00:17:21.010 I see a picture of you snowboarding on your background 370 00:17:21.010 --> 00:17:21.510 there. 371 00:17:21.510 --> 00:17:23.890 HELEN LEE: Yes, that's in Beijing. 372 00:17:23.890 --> 00:17:28.030 I used to do snowboarding when I was living in the D.C. area. 373 00:17:28.030 --> 00:17:30.242 SAM RANSBOTHAM: What was your first career that you 374 00:17:30.242 --> 00:17:31.450 wanted when you were a child? 375 00:17:31.450 --> 00:17:35.640 HELEN LEE: I would have to say mechanical engineer, 376 00:17:35.640 --> 00:17:37.740 because that was what my mom had been 377 00:17:37.740 --> 00:17:40.390 doing, because she designed household electrical 378 00:17:40.390 --> 00:17:41.340 appliances. 379 00:17:41.340 --> 00:17:45.730 I used to watch her draw those engineering drawings when 380 00:17:45.730 --> 00:17:48.000 I was little, and I thought, "Oh, that's amazing." 381 00:17:48.000 --> 00:17:51.910 And then you can see the product, and that's fun. 382 00:17:51.910 --> 00:17:54.860 So that's why my major in college 383 00:17:54.860 --> 00:17:58.120 was mechanical engineering, before I changed to aerospace 384 00:17:58.120 --> 00:17:58.970 engineering. 385 00:17:58.970 --> 00:18:01.420 SAM RANSBOTHAM: Both Shervin and I are chemical engineers, 386 00:18:01.420 --> 00:18:03.378 and so I have to bring that up every [episode]. 387 00:18:03.378 --> 00:18:05.540 And so we believe chemical engineering is better 388 00:18:05.540 --> 00:18:06.934 than all the other engineering. 389 00:18:06.934 --> 00:18:07.428 390 00:18:07.428 --> 00:18:09.970 HELEN LEE: [Laughs.] Well, yeah, in some of the universities, 391 00:18:09.970 --> 00:18:13.120 aerospace engineering is part of the mechanical engineering 392 00:18:13.120 --> 00:18:13.950 department. 393 00:18:13.950 --> 00:18:15.450 SAM RANSBOTHAM: What's your greatest 394 00:18:15.450 --> 00:18:16.960 wish for AI in the future? 395 00:18:16.960 --> 00:18:18.430 396 00:18:18.430 --> 00:18:20.180 HELEN LEE: I wish all the aircraft would 397 00:18:20.180 --> 00:18:23.940 be able to be controlled by AI, and also that the air traffic 398 00:18:23.940 --> 00:18:27.760 control would be conducted by AI so it's always 399 00:18:27.760 --> 00:18:31.580 just machine-to-machine talk, and so there would 400 00:18:31.580 --> 00:18:36.160 be fewer errors, less chance for mistakes, 401 00:18:36.160 --> 00:18:37.950 and [greater efficiency]. 402 00:18:37.950 --> 00:18:40.840 SAM RANSBOTHAM: I think we all want those things. 403 00:18:40.840 --> 00:18:43.320 Helen, it was great meeting you and talking with you. 404 00:18:43.320 --> 00:18:45.400 I think one thing that impressed me about this 405 00:18:45.400 --> 00:18:47.900 is what a complex environment you're in, 406 00:18:47.900 --> 00:18:50.380 coordinating lots of different organizations 407 00:18:50.380 --> 00:18:52.710 with equipment that's really out of your control. 408 00:18:52.710 --> 00:18:55.000 And it's a very difficult situation 409 00:18:55.000 --> 00:18:57.410 compared to a lot of the people who we talk to. 410 00:18:57.410 --> 00:18:59.340 Thank you for taking the time to talk with us. 411 00:18:59.340 --> 00:19:00.100 We really enjoyed it. 412 00:19:00.100 --> 00:19:00.570 Thanks. 413 00:19:00.570 --> 00:19:01.700 SHERVIN KHODABANDEH: Thank you very much. 414 00:19:01.700 --> 00:19:02.950 It's been really enlightening. 415 00:19:02.950 --> 00:19:04.840 HELEN LEE: Thank you for having me. 416 00:19:04.840 --> 00:19:07.590 SAM RANSBOTHAM: Next time, Shervin and I talk with Sowmya 417 00:19:07.590 --> 00:19:10.360 Gottipati, vice president of global supply chain 418 00:19:10.360 --> 00:19:12.650 technologies at the Estée Lauder Company. 419 00:19:12.650 --> 00:19:14.252 We hope you can join us. 420 00:19:14.252 --> 00:19:15.710 ALLISON RYDER: Thanks for listening 421 00:19:15.710 --> 00:19:17.170 to Me, Myself, and AI. 422 00:19:17.170 --> 00:19:19.670 We believe, like you, that the conversation 423 00:19:19.670 --> 00:19:21.890 about AI implementation doesn't start and stop 424 00:19:21.890 --> 00:19:23.007 with this podcast. 425 00:19:23.007 --> 00:19:24.840 That's why we've created a group on LinkedIn 426 00:19:24.840 --> 00:19:26.680 specifically for leaders like you. 427 00:19:26.680 --> 00:19:29.430 It's called AI for Leaders, and if you join us, 428 00:19:29.430 --> 00:19:31.440 you can chat with show creators and hosts, 429 00:19:31.440 --> 00:19:34.150 ask your own questions, share your insights, 430 00:19:34.150 --> 00:19:36.820 and gain access to valuable resources about AI 431 00:19:36.820 --> 00:19:39.650 implementation from MIT SMR and BCG. 432 00:19:39.650 --> 00:19:44.770 You can access it by visiting mitsmr.com/AIforLeaders. 433 00:19:44.770 --> 00:19:47.490 We'll put that link in the show notes, 434 00:19:47.490 --> 00:19:49.930 and we hope to see you there. 435 00:19:49.930 --> 00:19:55.000