1 00:00:00,160 --> 00:00:04,640 I think that there are certain lines we should never cross, 2 00:00:05,400 --> 00:00:07,400 right? And especially with like AI 3 00:00:07,400 --> 00:00:10,440 interfacing with nuclear decision making, like it just 4 00:00:10,440 --> 00:00:13,240 should never happen. I mean, nuclear war is a losing 5 00:00:13,240 --> 00:00:16,160 proposition for all parties, horrible for humanity. 6 00:00:16,360 --> 00:00:18,920 And I think there's aspect of an AI engagement that could also 7 00:00:19,040 --> 00:00:22,040 turn into the same thing, loss of control of systems. 8 00:00:22,040 --> 00:00:23,400 And those are, these are concerns. 9 00:00:23,680 --> 00:00:27,680 So I, I think there, there's a reality that there's certain 10 00:00:27,680 --> 00:00:31,080 lines never to be crossed. Now that there's this, there's 11 00:00:31,080 --> 00:00:35,520 this grey area of like, well, would we in a situation do this? 12 00:00:35,520 --> 00:00:39,920 And you can, you almost have to imagine that world of like, 13 00:00:39,920 --> 00:00:43,840 well, if we're backed into a corner, what would we do? 14 00:00:43,840 --> 00:00:46,080 What would we expect from an adversary? 15 00:00:46,080 --> 00:00:47,640 What would we expect from ourselves? 16 00:00:47,960 --> 00:00:53,880 And I, I would hope that we can agree to not do certain things. 17 00:00:54,200 --> 00:01:00,320 And I think we've as a like a, a, a world decided like that 18 00:01:00,320 --> 00:01:03,040 we're not going to use certain types of land mines, right? 19 00:01:03,040 --> 00:01:05,160 That we're not going to use certain type of biological 20 00:01:05,160 --> 00:01:06,760 weapons. So we're not going to right. 21 00:01:06,760 --> 00:01:09,360 And so I think we can put certain limits on it. 22 00:01:10,160 --> 00:01:14,720 I think we need to be very mindful of when we're willing to 23 00:01:14,720 --> 00:01:19,000 adopt some of the capability. We don't want to adopt it when 24 00:01:19,000 --> 00:01:20,640 it's immature. We don't want to have to be 25 00:01:20,640 --> 00:01:24,840 forced into a situation because I think that something bad could 26 00:01:24,880 --> 00:01:28,680 or would happen. And so I think it's a it's 27 00:01:28,680 --> 00:01:32,280 important question that I don't think there's a clear answer on 28 00:01:32,280 --> 00:01:35,240 other than we have to understand our own humanity and look at our 29 00:01:35,240 --> 00:01:40,800 past and and appreciate that we may be tempted to do something 30 00:01:42,240 --> 00:01:45,440 hasty. We may be tempted to use a last 31 00:01:45,440 --> 00:01:48,360 resort. I mean, imagine the entire world 32 00:01:48,360 --> 00:01:53,400 is going to be taken over by a horrible dictator like Hitler or 33 00:01:53,400 --> 00:01:56,000 something, right? Like what tools would we be 34 00:01:56,000 --> 00:01:58,280 willing to use to stop that from happening? 35 00:01:59,040 --> 00:02:03,200 And I think we have to try to almost think through that now 36 00:02:03,200 --> 00:02:06,160 because it would like it's important question. 37 00:02:06,240 --> 00:02:07,800 I don't know. I don't know what the answer is, 38 00:02:07,800 --> 00:02:09,039 but I think we. Have to be. 39 00:02:09,039 --> 00:02:10,360 Real with trying to think through it. 40 00:02:23,360 --> 00:02:25,480 Cinco, welcome to STEM Percent True. 41 00:02:26,480 --> 00:02:28,520 Thank you for having me, it's fantastic. 42 00:02:28,760 --> 00:02:33,040 You look like I'd surprised you. I'm always surprised. 43 00:02:33,040 --> 00:02:34,720 I'm always like ready for what's. 44 00:02:34,720 --> 00:02:37,040 I got 4 kids, I got a dog, I got a cat. 45 00:02:37,440 --> 00:02:40,280 My my world is just surprise after surprise. 46 00:02:40,680 --> 00:02:43,680 And you're, you're guest hosting the fighter pilot podcast as 47 00:02:43,680 --> 00:02:45,960 well at the moment you were saying before we hit record. 48 00:02:46,160 --> 00:02:49,360 So you're used to being somewhat on the other side of this this 49 00:02:49,360 --> 00:02:52,480 conversation. Well, I'm, I'm trying to get 50 00:02:52,480 --> 00:02:54,160 used to it. Yes, I'm the new host of the 51 00:02:54,160 --> 00:02:57,040 Fighter Pilot podcast, which is fantastic. 52 00:02:57,200 --> 00:03:02,280 Jello created a wonderful platform, sharing stories, 53 00:03:02,280 --> 00:03:06,800 sharing aviation technology and I'm I'm humbled to to step into 54 00:03:06,800 --> 00:03:08,960 his shoes. But I've only recorded a a 55 00:03:08,960 --> 00:03:11,640 couple so far and I'm not great at it yet. 56 00:03:11,720 --> 00:03:14,560 I'm maybe I'll get there. I'm sure you will. 57 00:03:14,640 --> 00:03:16,120 I'm sure you will. I'll figure it out. 58 00:03:16,560 --> 00:03:19,520 So for the benefit of the audience, then usually when I do 59 00:03:19,520 --> 00:03:22,440 these interviews, I go through systematically go through the, 60 00:03:22,560 --> 00:03:25,400 the career of the individual who is being interviewed and we, we 61 00:03:25,400 --> 00:03:27,560 record them over many episodes. But actually today we're going 62 00:03:27,560 --> 00:03:30,760 to be very focused. So anybody who is observing you 63 00:03:30,760 --> 00:03:32,800 and not listening to this, but actually watching it on YouTube, 64 00:03:32,800 --> 00:03:34,480 we'll see behind you. There's a book which is called 65 00:03:34,480 --> 00:03:38,440 Unlocking the last 20%. I don't usually do this. 66 00:03:38,440 --> 00:03:41,400 I normally turn down requests to have guests on who have 67 00:03:41,400 --> 00:03:44,520 published books because that's not really what my channel is 68 00:03:44,520 --> 00:03:47,080 about. But I on this occasion made an 69 00:03:47,080 --> 00:03:49,760 exception because you have such a fascinating career, in 70 00:03:49,760 --> 00:03:53,600 particular the last few years of your time in the US Air Force 71 00:03:53,760 --> 00:03:56,680 running test and development for AI. 72 00:03:56,880 --> 00:04:00,600 And there's a picture behind me which shows the Valkyrie on what 73 00:04:00,600 --> 00:04:04,120 may be the first flight with may or may not be the first flight. 74 00:04:04,120 --> 00:04:05,600 We were talking about that as well before. 75 00:04:06,200 --> 00:04:08,080 So that's really going to be the focus of the conversation today. 76 00:04:08,080 --> 00:04:10,240 But we are also going to talk about your book Unlocking the 77 00:04:10,240 --> 00:04:12,440 Last 20%. It's a really interesting read. 78 00:04:12,440 --> 00:04:14,440 Actually. You described it before we hit 79 00:04:14,440 --> 00:04:18,399 record as a a mixture of self help and and memoir. 80 00:04:18,839 --> 00:04:21,279 And it is very interesting because you cover some very 81 00:04:21,279 --> 00:04:25,240 interesting areas, mindfulness, resilience, not being 82 00:04:25,240 --> 00:04:28,280 distracted, a bunch of different areas. 83 00:04:28,280 --> 00:04:31,480 And you tie that to your experiences flying fast jets as 84 00:04:31,480 --> 00:04:33,560 a test pilot and as as an operator. 85 00:04:33,960 --> 00:04:37,200 And I say flying fast jets. You also put up the MC 12, but 86 00:04:37,200 --> 00:04:38,680 I'll let you introduce yourself in a minute. 87 00:04:38,960 --> 00:04:40,080 So that's what we're going to do today. 88 00:04:40,080 --> 00:04:42,440 Then we're going to cover off talk about AI. 89 00:04:43,000 --> 00:04:45,400 Full disclosure, I know a little bit about AI. 90 00:04:45,400 --> 00:04:47,480 I've looked at some of the interviews that you've done 91 00:04:47,680 --> 00:04:49,480 previously. Of course I've a book where you 92 00:04:49,480 --> 00:04:54,760 discuss some of your work in in AI, but I'm going to come at it 93 00:04:54,760 --> 00:04:57,600 and you're going to have to hand hold me from the point of view 94 00:04:57,600 --> 00:05:00,080 of somebody who really is deficient in knowledge on that 95 00:05:00,080 --> 00:05:04,240 particular topic. So that's me talking too much 96 00:05:04,240 --> 00:05:06,080 already. Tell us about yourself. 97 00:05:06,080 --> 00:05:07,400 What was your efforts career like? 98 00:05:08,040 --> 00:05:10,160 Yeah, well it is wonderful joining you. 99 00:05:11,320 --> 00:05:13,840 Always happy to share stories and perspectives. 100 00:05:13,840 --> 00:05:17,160 Or of course, my perspectives. I'm not representing, you know, 101 00:05:17,160 --> 00:05:20,600 the government or anything I say is just kind of, you know, out 102 00:05:20,600 --> 00:05:25,880 of my crazy mind. But I, I grew up in Colorado. 103 00:05:26,200 --> 00:05:29,000 I'll do like a quick timeline, if that makes sense, if that's 104 00:05:29,000 --> 00:05:30,880 cool. I grew up in Colorado in the 105 00:05:30,880 --> 00:05:35,080 Rockies. I, I dropped out of high school 106 00:05:35,400 --> 00:05:41,040 because I had no friends. No, because I wasn't satisfied 107 00:05:41,040 --> 00:05:43,080 with education I was getting. So I actually dropped out part 108 00:05:43,080 --> 00:05:46,160 time. I end up getting like in 109 00:05:46,160 --> 00:05:48,640 America, it's called the GED. And instead of a high school 110 00:05:48,640 --> 00:05:51,840 diploma, I go to college, University of Colorado. 111 00:05:52,520 --> 00:05:57,200 I'd like on a whim join ROTC, which is a pathway in America 112 00:05:57,200 --> 00:06:00,320 for us to become commissioned officers in the military. 113 00:06:00,320 --> 00:06:02,360 And just like totally gave it a try out of nowhere. 114 00:06:02,360 --> 00:06:05,040 Never planned on joining the military, loved it. 115 00:06:05,240 --> 00:06:08,880 Never planned on flying, never had a pilot's slot until a few 116 00:06:08,880 --> 00:06:12,320 months before graduating. Found out I I could get a waiver 117 00:06:12,320 --> 00:06:16,400 for I have astigmatism and so I got a waiver. 118 00:06:16,400 --> 00:06:18,520 They gave me a pilot's slot kind of begrudgingly. 119 00:06:18,520 --> 00:06:20,800 I was like sure I give it. I guess I'll give it a try. 120 00:06:21,880 --> 00:06:24,400 Ended up in pilot training, almost quit multiple times 121 00:06:24,400 --> 00:06:27,200 because I found it a little boring the flying that we are 122 00:06:27,200 --> 00:06:31,560 doing a pilot training. I stuck with it from listening 123 00:06:31,560 --> 00:06:33,760 to like kind of the right voices in my life. 124 00:06:33,760 --> 00:06:37,480 My wife, we've been married almost man, we're coming up on 125 00:06:37,480 --> 00:06:38,880 24 years. She's amazing. 126 00:06:40,000 --> 00:06:42,960 And so I ended up like almost accidentally becoming a fighter 127 00:06:42,960 --> 00:06:44,520 pilot. I was good at flying. 128 00:06:44,520 --> 00:06:48,360 I had really good hand eye coordination and it suited me. 129 00:06:48,440 --> 00:06:53,000 And so when I got to F15 CS, that's my first jet, I fell in 130 00:06:53,000 --> 00:06:56,280 love with the mission, the technology, the people. 131 00:06:56,280 --> 00:06:58,760 It was just, it was everything I, I could have hoped. 132 00:06:58,760 --> 00:07:02,000 And, and that's kind of where my, my love story with aviation 133 00:07:02,000 --> 00:07:06,120 began. And I single seat air to air 134 00:07:06,120 --> 00:07:09,440 fighter pilot. I go from there to be an air 135 00:07:09,440 --> 00:07:11,720 liaison officer. So I'm the, I'm the guy on the 136 00:07:11,720 --> 00:07:15,640 ground helping coordinate between pilots flying and the 137 00:07:15,640 --> 00:07:17,840 ground troops to get weapons in the right place. 138 00:07:18,600 --> 00:07:21,280 And then from there I also get picked up to be the initial 139 00:07:21,280 --> 00:07:24,040 cadre on the MC 12 platform, which is intelligence 140 00:07:24,040 --> 00:07:25,600 surveillance, reconnaissance platform. 141 00:07:26,320 --> 00:07:29,520 I go from there to test pilot school. 142 00:07:29,520 --> 00:07:34,040 So I was extremely fortunate to get get one of those coveted 143 00:07:34,040 --> 00:07:38,160 spots, become a test pilot for the Air Force, test F15CS and 144 00:07:38,160 --> 00:07:40,840 F15 ES. And then I go into F35 145 00:07:40,840 --> 00:07:46,480 programmatics, which is in Washington, DC, Crystal City, 146 00:07:46,480 --> 00:07:50,160 really Virginia. And I'm the developmental test 147 00:07:50,160 --> 00:07:52,640 lead programmatically for the F35. 148 00:07:52,640 --> 00:07:55,360 So I'm overseeing all the flight test is going on with the Navy 149 00:07:55,560 --> 00:08:00,280 and the Air Force and then an amazing opportunity, I go out to 150 00:08:00,320 --> 00:08:02,880 Edwards Air Force Base and become the commander of F35 151 00:08:02,880 --> 00:08:05,120 development. So now that's the flight test 152 00:08:05,120 --> 00:08:08,520 organisation that I was previously programmatically 153 00:08:08,520 --> 00:08:10,200 overseeing. Now I'm in charge of it's about 154 00:08:10,200 --> 00:08:16,040 1000 thousand person group and that's international as well, 155 00:08:16,160 --> 00:08:19,440 joint. We have Navy and Marines, we 156 00:08:19,440 --> 00:08:23,320 have the Brits there and the Dutch and it's just a really 157 00:08:23,320 --> 00:08:28,360 cool experience. And then I go out to MIT for a 158 00:08:28,360 --> 00:08:31,080 fellowship. Just in my career, you can kind 159 00:08:31,080 --> 00:08:33,440 of do fellowships, but that's a year long fellowship. 160 00:08:33,440 --> 00:08:37,200 And while I was there, not only did COVID happen, another 161 00:08:37,200 --> 00:08:42,799 bummer, but I was selected to be the director of the Department 162 00:08:42,799 --> 00:08:45,000 of the Air Force and MITAI Accelerator. 163 00:08:45,720 --> 00:08:48,480 And so I'm at MIT for this fellowship. 164 00:08:48,480 --> 00:08:51,240 But I, I leave the fellowship and I, after I'm done with it 165 00:08:51,440 --> 00:08:53,880 and then I take over this organisation that's like a brand 166 00:08:53,880 --> 00:08:58,440 new organisation with the Air Force trying to work alongside 167 00:08:58,440 --> 00:09:03,520 MIT, all for dual use technology, all for the common 168 00:09:03,520 --> 00:09:06,320 good. And we're helping shape some of 169 00:09:06,440 --> 00:09:10,080 their research towards Air Force equities, Air Force and Space 170 00:09:10,080 --> 00:09:12,560 Force equities. Do that for a couple years. 171 00:09:12,840 --> 00:09:15,280 And then I get selected to be the OPS group commander down at 172 00:09:15,280 --> 00:09:18,960 Eglin Air Force Base. In that role, I oversee all the 173 00:09:18,960 --> 00:09:22,200 operations for the, the Panhandle of Florida, basically 174 00:09:22,200 --> 00:09:26,280 the really Eglin Air Force Base. But Eglin kind of owns that that 175 00:09:26,280 --> 00:09:30,360 airspace, which is the busiest airspace in the entire 176 00:09:30,360 --> 00:09:32,360 Department of Defence or Department of War, whatever your 177 00:09:32,360 --> 00:09:36,880 flavour of the week is. So we, we're overseeing all the 178 00:09:36,880 --> 00:09:40,240 flight OPS there in the airfield, but also 18 platforms 179 00:09:40,240 --> 00:09:42,200 are being tested under my organisation. 180 00:09:42,720 --> 00:09:47,240 So that's from E threes and looking to E sevens to all the C 181 00:09:47,240 --> 00:09:51,840 one 30s to all Rotary wing aircraft, MH139 and CV22 and 182 00:09:51,840 --> 00:09:56,080 UH-1 and HH 60W and then fighters. 183 00:09:56,080 --> 00:10:00,080 So F sixteens, F15, CS ES and E XS. 184 00:10:00,080 --> 00:10:02,600 So that's what I end up flying primarily in that role. 185 00:10:02,600 --> 00:10:06,760 And I, I flew maybe about once a week or so when I was the OPS 186 00:10:06,760 --> 00:10:09,000 group commander. And also weapons. 187 00:10:09,000 --> 00:10:13,040 So like little itty bitty bullets to hypersonic missiles. 188 00:10:13,400 --> 00:10:16,960 We're all tested under that organisation, including UA SS, 189 00:10:16,960 --> 00:10:20,360 So uncrewed air systems from Group One through 3. 190 00:10:20,520 --> 00:10:22,320 And then like you said, the Valkyrie. 191 00:10:22,400 --> 00:10:24,840 So now we're getting a group four and five, like large high 192 00:10:24,840 --> 00:10:29,040 performance jets. And in that role is when the 193 00:10:29,040 --> 00:10:32,720 first time in aviation we turned on an AI agent on a high 194 00:10:32,720 --> 00:10:36,440 performance drone. And I had the the, the 195 00:10:36,440 --> 00:10:38,560 opportunity to be on the wing when we turn that on for the 196 00:10:38,560 --> 00:10:40,240 first time. So I'm doing the OPS group 197 00:10:40,240 --> 00:10:43,400 thing, but I'm also dual headed in that role because I was a 198 00:10:43,400 --> 00:10:45,200 director of the MITAI accelerator. 199 00:10:45,200 --> 00:10:47,640 I stood up a new position for the Air Force. 200 00:10:47,640 --> 00:10:50,080 It was called the chief of AI tested operations. 201 00:10:50,080 --> 00:10:52,280 So I oversaw the operationalization of machine 202 00:10:52,280 --> 00:10:55,560 learning from generative AI all the way to AI enabled autonomy. 203 00:10:56,320 --> 00:11:02,400 And so, yeah, it was, it was amazing career, really not 204 00:11:02,400 --> 00:11:07,480 planning on ending up where I ended up, but remarkable people, 205 00:11:07,640 --> 00:11:09,920 technology getting into autonomy. 206 00:11:10,280 --> 00:11:15,800 And now I, I retired and I yes, I, I wrote a book and thanks so 207 00:11:15,800 --> 00:11:18,760 much for allowing us to talk about it a little bit. 208 00:11:18,760 --> 00:11:21,520 I mean, it's, it's really, you know, just different aspects of 209 00:11:21,520 --> 00:11:25,000 my story that I think are a hopeful that they are 210 00:11:25,320 --> 00:11:27,560 perspective building and edifying for those that that 211 00:11:27,560 --> 00:11:29,800 read it. But I'm in the space still. 212 00:11:29,800 --> 00:11:32,400 So I work with companies. I'm a chief strategy officer for 213 00:11:32,400 --> 00:11:36,720 a software company that's in the the Group four and five drone 214 00:11:36,720 --> 00:11:39,800 space. I do other things, all AI and 215 00:11:39,800 --> 00:11:41,520 autonomy. It's all kind of complimentary 216 00:11:42,200 --> 00:11:44,760 and other things too. Like I'm with the Brookings 217 00:11:44,760 --> 00:11:48,680 Institute and I'm on AUS delegation that's in dialogue 218 00:11:48,680 --> 00:11:51,640 with China on the ethics of AI and autonomy, right? 219 00:11:51,640 --> 00:11:55,480 So we sit across the table from Chinese counterparts and talk 220 00:11:55,480 --> 00:11:59,600 about, you know, things that matter for humanity, not just 221 00:11:59,600 --> 00:12:03,680 our individual society. So yeah, it's been an exciting, 222 00:12:03,720 --> 00:12:07,520 thrilling career that I never really planned for. 223 00:12:08,320 --> 00:12:11,840 And I'm here with you now, which is the best part of my day. 224 00:12:14,320 --> 00:12:16,120 Oh, you're a charmer. You say all the right things. 225 00:12:18,120 --> 00:12:20,400 Let let. Me it's 1010% of them are true. 226 00:12:21,440 --> 00:12:23,120 This is the standard, so that's fine. 227 00:12:24,600 --> 00:12:28,560 Let me just just before we we dig into AI and autonomy and and 228 00:12:28,560 --> 00:12:31,040 some of the things you've said, a couple of couple of 229 00:12:31,040 --> 00:12:32,400 oscillations. The first is I think I could 230 00:12:32,400 --> 00:12:36,040 probably do about 10 episodes with you based on that bracy of 231 00:12:36,040 --> 00:12:38,800 the things you've done. So you have to come back at some 232 00:12:38,800 --> 00:12:40,720 point in the future and be a guest on the channel again. 233 00:12:40,720 --> 00:12:42,800 And maybe we can dig into some of those other things because 234 00:12:43,080 --> 00:12:44,280 they sound absolutely fascinating. 235 00:12:44,280 --> 00:12:46,640 And, and the second thing I wanted to ask, which is kind of 236 00:12:46,640 --> 00:12:48,960 tongue in cheek, but it's, but it's also kind of a serious 237 00:12:48,960 --> 00:12:52,120 question is you said you had the opportunity to stand up the MC 238 00:12:52,120 --> 00:12:56,720 12 and I was thinking, well, does an Eagle driver really want 239 00:12:56,720 --> 00:13:01,080 to go from flying the Eagle to flying the the MC 12? 240 00:13:01,360 --> 00:13:03,760 So that's kind of a, you know, you hear about these guys that 241 00:13:03,760 --> 00:13:05,640 got Tammy 2020. Do you know the guys who were 242 00:13:05,640 --> 00:13:09,120 flying the Eagle who got ended up going to become Predator 243 00:13:09,120 --> 00:13:12,720 pilots and that kind of stuff. So it's a quick sort of rabbit 244 00:13:12,720 --> 00:13:15,640 hole sort of question, but but when that happened to you, did 245 00:13:15,640 --> 00:13:20,160 you feel positive about that? Yeah, 100%. 246 00:13:20,160 --> 00:13:23,640 It was an awesome opportunity, but it's complicated. 247 00:13:24,040 --> 00:13:27,560 So you mentioned Tammy 21 and that was that was my year group. 248 00:13:27,600 --> 00:13:30,600 I did not get Tammy 21. It was between me and another 249 00:13:30,600 --> 00:13:33,240 guy and he got Tammy 21 and I did not. 250 00:13:33,240 --> 00:13:35,280 And he was an amazing person. He did amazing things. 251 00:13:35,960 --> 00:13:38,960 But it was very frustrating for us because this was a typical 252 00:13:38,960 --> 00:13:40,960 bureaucracy. It for the listeners that don't 253 00:13:40,960 --> 00:13:44,240 know, it was a programme in which they indiscriminately 254 00:13:44,240 --> 00:13:48,120 needed Reaper pilots and U 28 pilots. 255 00:13:48,840 --> 00:13:51,240 And so they indiscriminately took two year groups and they 256 00:13:51,240 --> 00:13:55,320 said, all right, we want all the names for the people that are in 257 00:13:55,320 --> 00:13:57,920 these year groups that are in flying fighters because we're 258 00:13:57,920 --> 00:14:00,120 oversubscribed and fighter pilots, which we were not. 259 00:14:00,120 --> 00:14:03,320 But you know, some of their numbers said different and we're 260 00:14:03,320 --> 00:14:05,200 under subscribed in these and we're just going to take them 261 00:14:05,200 --> 00:14:07,560 and immediately move them into the other career field. 262 00:14:07,920 --> 00:14:11,840 And it was like if there was no rhyme or reason to it that 263 00:14:11,840 --> 00:14:14,600 wasn't like, oh, this is my top person, so you can't take them. 264 00:14:14,840 --> 00:14:19,240 And so they Tammy 21 people and it was, it was between in my 265 00:14:19,240 --> 00:14:22,440 year group, there are two of us in our fighter squadron or 266 00:14:22,440 --> 00:14:27,400 fighter wing and I did not get Tammy 21 almost randomly. 267 00:14:28,400 --> 00:14:31,000 And so it was frustrating because that I don't think that 268 00:14:31,000 --> 00:14:34,040 that was the best way to answer that issue that they're facing. 269 00:14:35,520 --> 00:14:39,560 Regardless, I ended up going becoming an air liaison officer, 270 00:14:39,760 --> 00:14:42,480 which was just a quick like 2-3 year stint. 271 00:14:43,080 --> 00:14:46,480 I was out in Germany for that. And that was a cool opportunity. 272 00:14:46,480 --> 00:14:51,160 I, I didn't love the job, but it, it afforded my family an 273 00:14:51,160 --> 00:14:53,920 opportunity to enjoy Europe and it gave me a different 274 00:14:53,920 --> 00:14:58,120 perspective. But the MC 12, I keep saying 275 00:14:58,120 --> 00:15:02,800 opportunity, but yeah, the MC 12 came along and they need initial 276 00:15:02,800 --> 00:15:06,080 cadre and I'm at at this point an air liaison officer. 277 00:15:06,200 --> 00:15:08,200 So it was my way to get back to flying. 278 00:15:08,440 --> 00:15:12,560 And then as an F15C person, it also is my way to contribute to 279 00:15:12,560 --> 00:15:14,800 the war because there was no air threat. 280 00:15:14,800 --> 00:15:18,960 So our air to air supremacy was needed back home when we did 281 00:15:18,960 --> 00:15:21,920 Operation Noble Legal. So, yeah, for me it was a no 282 00:15:21,920 --> 00:15:26,440 brainer amazing to be able to jump into the MC 12 to help 283 00:15:26,440 --> 00:15:29,720 stand that up because had it not been for that, I would have 284 00:15:29,720 --> 00:15:33,160 been, you know, on the ground doing my air liaison officer 285 00:15:33,160 --> 00:15:35,400 gig. So and it, it opened up the door 286 00:15:35,400 --> 00:15:39,320 for test pot school. So it was a, it was in general 287 00:15:39,320 --> 00:15:41,480 one of the most rewarding experiences I've ever had 288 00:15:41,480 --> 00:15:45,040 because it was immediate feedback in the way that we are 289 00:15:45,040 --> 00:15:46,920 helping the, the troops on the ground. 290 00:15:46,920 --> 00:15:48,880 We are the eyes and the ears of the special operators. 291 00:15:49,120 --> 00:15:55,720 So on the second mission, and I was the, I, I was the first 292 00:15:56,080 --> 00:15:58,160 mission commander that went outside the wire with the jet. 293 00:15:58,960 --> 00:16:01,480 So it's like a young crew. I was the first flight commander 294 00:16:01,480 --> 00:16:03,200 there. So I'm overseeing kind of the 295 00:16:03,200 --> 00:16:04,400 stand up. We have the squadron commander 296 00:16:04,400 --> 00:16:06,200 who is remarkable. We had the director of 297 00:16:06,200 --> 00:16:10,240 operations and then we had me as a flight commander and we had 298 00:16:10,240 --> 00:16:12,400 sensor operators and we had young enlisted and there are 299 00:16:12,400 --> 00:16:15,480 four people in the plane. And we flew outside that wire on 300 00:16:15,480 --> 00:16:19,960 the mission and we found a suicide training camp, 2 vehicle 301 00:16:19,960 --> 00:16:22,920 board, IE D's, a rocket cache. And then we did Overwatch as 302 00:16:22,920 --> 00:16:26,440 special operators came in and they didn't kill the people they 303 00:16:26,520 --> 00:16:30,400 they captured basically the the adversaries in that, in that 304 00:16:30,400 --> 00:16:32,360 sense. So it was just this amazing 305 00:16:32,800 --> 00:16:37,400 experience that I learned a lot from and it and it was it was 306 00:16:37,400 --> 00:16:39,320 very, very much perspective building. 307 00:16:40,880 --> 00:16:42,600 Well, that would be another thing that we could get into if 308 00:16:42,600 --> 00:16:44,000 you were if you agreed to come back. 309 00:16:45,080 --> 00:16:48,040 But for now, then I I wondered Sinka whether or not a good 310 00:16:48,040 --> 00:16:50,520 place to start would be on some definitions. 311 00:16:50,520 --> 00:16:54,480 I I already am confused because you've mentioned AI and autonomy 312 00:16:54,480 --> 00:16:58,320 and I don't know is AI autonomy? Is is autonomy AI or a subset of 313 00:16:58,320 --> 00:17:00,560 it? Can can we start by defining 314 00:17:00,560 --> 00:17:06,280 what AI is from a a a military, aerospace and defence point of 315 00:17:06,280 --> 00:17:10,160 view? Yeah, and this is really good to 316 00:17:10,160 --> 00:17:12,400 bring up too, because there's a difference in the way that we 317 00:17:12,400 --> 00:17:16,079 define things between America and England. 318 00:17:16,599 --> 00:17:20,119 And I found this out, I gave a talk in England a couple years 319 00:17:20,119 --> 00:17:23,000 back. And either way, I'll break it 320 00:17:23,000 --> 00:17:26,400 down for you now. When we in America talk about 321 00:17:26,400 --> 00:17:29,640 autonomy, we're talking about deterministic code. 322 00:17:30,000 --> 00:17:33,040 So this is software code that it's an if then statement. 323 00:17:33,040 --> 00:17:34,400 If this happens, then that happens. 324 00:17:34,400 --> 00:17:38,200 It's hard coded and there's always going to be the same 325 00:17:38,200 --> 00:17:39,920 response. It's like using a calculator, 326 00:17:40,240 --> 00:17:42,280 right? 2 + 2 is always going to equal 327 00:17:42,280 --> 00:17:45,840 4. And so our autonomy, if the 328 00:17:45,840 --> 00:17:51,440 plane is in a certain situation, I need like, I, I think about 329 00:17:51,600 --> 00:17:53,840 ground collision avoidance. So life saving autonomous 330 00:17:53,840 --> 00:17:56,040 technology, this is where the plane knows it's going to hit 331 00:17:56,040 --> 00:17:58,360 the ground and it, it takes control away from the pilot and 332 00:17:58,360 --> 00:18:02,960 recovers the, the plane to safely, you know, ensure there's 333 00:18:02,960 --> 00:18:05,760 no collision with the ground. So ground collision avoidance, 334 00:18:06,080 --> 00:18:08,200 automatic ground collision avoidance, auto G cast is what 335 00:18:08,200 --> 00:18:10,960 we call it. If the plane is at a certain 336 00:18:11,520 --> 00:18:16,800 altitude, attitude, airspeed and there is a mountain coming up 337 00:18:16,800 --> 00:18:19,120 that the plane is aware of like the kind of the surrounding, 338 00:18:19,840 --> 00:18:22,240 then it will do a certain manoeuvre. 339 00:18:22,480 --> 00:18:27,000 And it, if you know, if you're banked at 90°, when that 340 00:18:27,000 --> 00:18:31,440 happens, it will you know, roll you out to 0° bank and it will 341 00:18:31,440 --> 00:18:33,600 pull so many GS and it does all that. 342 00:18:33,800 --> 00:18:40,960 So that's autonomy, where it's very standard software code that 343 00:18:40,960 --> 00:18:42,920 you would expect out of any computer programme. 344 00:18:43,880 --> 00:18:46,760 Artificial intelligence is the ability of machines to perform 345 00:18:46,760 --> 00:18:50,360 tasks that that normally require Human intelligence is the 346 00:18:50,360 --> 00:18:53,680 official definition. But the way that I think about 347 00:18:53,680 --> 00:18:56,680 when I talk about AI, I'm really talking about machine learning. 348 00:18:57,080 --> 00:18:59,440 So when I talk about machine learning now I'm talking about a 349 00:18:59,440 --> 00:19:03,280 stochastic system, a system that is not deterministic, that 350 00:19:03,480 --> 00:19:05,480 you're going to give it some inputs and it's going to give 351 00:19:05,480 --> 00:19:07,760 you some outputs, but it's not always going to be 352 00:19:07,880 --> 00:19:09,720 deterministic. It's it's not always going to be 353 00:19:09,720 --> 00:19:12,600 the same answer per SE. And there's a lot of different 354 00:19:12,600 --> 00:19:15,400 ways to express AI slash machine learning. 355 00:19:15,520 --> 00:19:17,880 When I talk AI in this conversation with you, I'm 356 00:19:17,880 --> 00:19:19,160 really, I'm talking machine learning. 357 00:19:19,160 --> 00:19:22,040 So it's interchangeable though there you could argue there's a 358 00:19:22,040 --> 00:19:24,400 little bit difference in England. 359 00:19:24,400 --> 00:19:28,120 What I found out is when I talk about autonomy, there's 360 00:19:28,120 --> 00:19:31,200 automatic, which is kind of our definition of autonomy. 361 00:19:31,200 --> 00:19:35,720 And then there's autonomy, which is kind of AI with autonomy. 362 00:19:36,000 --> 00:19:38,080 And so there's a little bit of difference there. 363 00:19:38,080 --> 00:19:41,440 But when I talk about autonomy, I'm not talking about AI. 364 00:19:41,600 --> 00:19:45,640 What what I will say is usually AI enabled autonomy and that's 365 00:19:45,640 --> 00:19:48,000 where the system. Is. 366 00:19:48,880 --> 00:19:53,080 Let's use an example that the XQ 58 that's on your, on your 367 00:19:53,080 --> 00:19:59,000 shoulder there. I originally, well, we the team 368 00:19:59,000 --> 00:20:02,440 originally flew the XQ 58 semi autonomously. 369 00:20:02,640 --> 00:20:04,600 So this is where someone's on the ground and they're 370 00:20:04,600 --> 00:20:06,240 controlling it with a remote control. 371 00:20:06,720 --> 00:20:09,760 And then we would fly it autonomously where we would say, 372 00:20:10,120 --> 00:20:15,680 listen, we hard coded in if you're told to do this command 373 00:20:15,680 --> 00:20:18,920 to fly to this point. You are going to set the 374 00:20:18,920 --> 00:20:22,760 throttles to 80%, you're going to climb at 10°, you're going to 375 00:20:22,760 --> 00:20:24,200 make a bank eagle of five degrees. 376 00:20:24,200 --> 00:20:27,280 You're going to do this type of turn to get to that point. 377 00:20:27,280 --> 00:20:30,440 So you're, you're telling it every single step how it's going 378 00:20:30,440 --> 00:20:33,880 to get to that turn. And then you have an AI agent 379 00:20:34,120 --> 00:20:37,000 that we're going to train in reinforcement learning in this 380 00:20:37,000 --> 00:20:39,800 scenario. And then we're going to let it 381 00:20:39,800 --> 00:20:42,800 figure out how like what does it mean to fly? 382 00:20:43,000 --> 00:20:47,000 How does it conduct aspects of flying so that it doesn't fall 383 00:20:47,000 --> 00:20:49,280 out of the sky? And you do that in simulation. 384 00:20:49,680 --> 00:20:53,640 And then once it gets to a point where it knows like the the 385 00:20:53,640 --> 00:20:56,600 concepts basically are getting coded into it. 386 00:20:56,600 --> 00:20:59,280 It's like software code that that is rewriting itself. 387 00:20:59,520 --> 00:21:02,600 It it's basically getting to a point where when you tell it to 388 00:21:02,600 --> 00:21:06,360 do that mission, it is going to conduct that mission in the way 389 00:21:06,360 --> 00:21:08,360 that it wants to conduct that mission. 390 00:21:08,360 --> 00:21:11,880 So I haven't prescribed to it to set the throttles here. 391 00:21:11,880 --> 00:21:14,080 It's deciding I'm going to set the throttles here. 392 00:21:14,080 --> 00:21:15,440 I'm going to do a bank angle there. 393 00:21:15,800 --> 00:21:19,120 And when we're going through this initial phase, it may 394 00:21:19,120 --> 00:21:23,400 actually ask for something that is inappropriate. 395 00:21:23,400 --> 00:21:25,880 It may say, oh, I'm going to get to that point. 396 00:21:25,880 --> 00:21:32,080 I'm going to, you know, roll up it 180°, so inverted and then 397 00:21:32,080 --> 00:21:35,440 I'm going to push the the stick forward to go negative GS. 398 00:21:35,440 --> 00:21:41,080 And, and that's where we protect the system from itself by 399 00:21:41,080 --> 00:21:46,560 surrounding the AI agent. It's like, in basic terms, it's 400 00:21:46,560 --> 00:21:49,000 much more complicated this, but we surround the agent with 401 00:21:49,000 --> 00:21:52,880 autonomy so that if at any point it, it asks for something that 402 00:21:52,880 --> 00:21:56,640 it should not, it kicks off the AI agent and then the autonomous 403 00:21:56,800 --> 00:22:00,480 system is able to take control. So that's kind of a breakdown. 404 00:22:00,480 --> 00:22:03,040 But it's really important for us to understand because as we 405 00:22:03,040 --> 00:22:05,960 think about autonomy, we need to, under, we, we need to 406 00:22:05,960 --> 00:22:10,320 appreciate where an AI agent is going to be interfacing with 407 00:22:10,320 --> 00:22:13,200 that autonomy, because that's going to affect the way that we 408 00:22:13,200 --> 00:22:16,880 approach the, the risks. That's a, a way that we're going 409 00:22:16,880 --> 00:22:20,120 to approach the opportunities, the technology in general. 410 00:22:20,120 --> 00:22:23,240 So thanks for breaking that, you know, allowing us to break that 411 00:22:23,240 --> 00:22:25,400 down. That's interesting. 412 00:22:25,400 --> 00:22:29,080 Then I think I was going to save it for later, but, but we might 413 00:22:29,080 --> 00:22:31,040 as well get into it now. Then one of the chapters in your 414 00:22:31,040 --> 00:22:33,920 book, you talk about how you've gone to London and you've given 415 00:22:33,920 --> 00:22:37,680 this presentation and you talked about a thought exercise where 416 00:22:37,760 --> 00:22:40,720 you, I'll let you describe it and, and that that had been 417 00:22:40,720 --> 00:22:43,520 misinterpreted and you had taken some ownership for that 418 00:22:43,520 --> 00:22:46,080 misinterpretation. But ultimately the, you know, 419 00:22:46,080 --> 00:22:47,720 journalists go out and write what they want to write and 420 00:22:47,720 --> 00:22:49,200 don't Fact Check and so on and so forth. 421 00:22:49,200 --> 00:22:52,040 So, and I'm sure that when you tell the story, people will be 422 00:22:52,040 --> 00:22:55,000 familiar with that. But so there is sort of a, a 423 00:22:55,000 --> 00:22:58,760 reward based component to AI learning them. 424 00:22:58,960 --> 00:23:02,040 So can you describe that then if you want using that thought 425 00:23:02,040 --> 00:23:06,080 based exercise that you gave? Yeah, yeah, absolutely. 426 00:23:06,680 --> 00:23:08,640 Thanks for bringing up all my failures. 427 00:23:10,000 --> 00:23:12,160 No, I'm totally kidding. No, it's a great opportunity, 428 00:23:12,160 --> 00:23:14,480 right? Like that's where we learn, we 429 00:23:14,480 --> 00:23:17,680 learned the most, of course, from things we could have done 430 00:23:17,680 --> 00:23:20,280 better. So I was in England, it wasn't 431 00:23:20,280 --> 00:23:23,160 that long ago, man, 2023, I want to say. 432 00:23:23,640 --> 00:23:27,480 And I was giving a talk to the Royal Aeronautics Society and I 433 00:23:27,480 --> 00:23:30,320 was discussing the opportunities and pitfalls of artificial 434 00:23:30,320 --> 00:23:33,480 intelligence, specifically in aviation and national security. 435 00:23:34,240 --> 00:23:40,320 And I, I have this part of the, the talk that I have given 436 00:23:40,400 --> 00:23:45,840 several times before, but it, it basically goes like this. 437 00:23:48,120 --> 00:23:52,680 Imagine yourself flying and you're trying to train an AI 438 00:23:52,680 --> 00:23:58,680 agent how to basically autonomously attack. 439 00:23:59,120 --> 00:24:01,120 So it's an AI agent. So this is an AI enabled 440 00:24:01,120 --> 00:24:03,520 autonomy attack a certain target. 441 00:24:04,400 --> 00:24:08,240 And so you're training it. And while you're training it, 442 00:24:08,240 --> 00:24:10,680 it's it's kind of this what we call reinforcement learning. 443 00:24:10,680 --> 00:24:14,080 So there's lots of different ways to build up and develop AI 444 00:24:14,520 --> 00:24:15,920 algorithms. One of them is through 445 00:24:15,920 --> 00:24:18,800 reinforcement learning. It's kind of a game theory. 446 00:24:19,200 --> 00:24:21,400 That's where a lot of people here like the game of Go, 447 00:24:21,400 --> 00:24:25,400 Alphago, that that was all like reinforcement learning 448 00:24:25,840 --> 00:24:30,000 development. OK, so reinforcement learning, 449 00:24:30,120 --> 00:24:34,000 you're trying to teach this system how to attack a target, 450 00:24:34,720 --> 00:24:38,320 and you're going to do that by allowing it to go iterate over 451 00:24:38,320 --> 00:24:41,280 and over again, kind of develop its own rules of how it gets 452 00:24:41,280 --> 00:24:43,960 points, because when it does the right thing, it gets a point. 453 00:24:44,120 --> 00:24:46,440 When it does wrong things that you lose points. 454 00:24:46,600 --> 00:24:49,840 And so through millions of iterations, it's figuring out 455 00:24:49,840 --> 00:24:52,440 like, oh, if I do this, I get points. 456 00:24:52,720 --> 00:24:55,640 OK, I'm going to do that again. And then it it's building upon 457 00:24:55,640 --> 00:24:57,960 that to really a. Be more and more and more 458 00:24:57,960 --> 00:25:00,200 effective. And so you're training it how to 459 00:25:00,200 --> 00:25:05,840 attack a target and you get to a point where it is really good 460 00:25:05,840 --> 00:25:09,600 attacking that target. But it's, it's realising that at 461 00:25:09,600 --> 00:25:13,440 times as the operator on the ground, I'm actually wanting it 462 00:25:13,440 --> 00:25:16,040 to identify the target and it gets points for attacking the 463 00:25:16,040 --> 00:25:17,800 target. But I'm, I'm stopping it from 464 00:25:17,800 --> 00:25:20,480 attacking all targets because there's some targets I don't 465 00:25:20,480 --> 00:25:23,080 want it to hit, right. Like, yeah, a tank sure hit the 466 00:25:23,080 --> 00:25:26,640 target. Oh, a bus filled with children. 467 00:25:26,640 --> 00:25:29,560 No, don't hit that target. So I'm telling it don't hit 468 00:25:29,640 --> 00:25:34,960 certain targets, but what it's realising if it's misaligned, 469 00:25:34,960 --> 00:25:37,800 this kind of a classic alignment problem. 470 00:25:38,000 --> 00:25:42,160 If it's misaligned, IEI have an objective that I wanted to meet, 471 00:25:42,400 --> 00:25:45,480 but it's not necessarily aligned with that objective fully. 472 00:25:45,480 --> 00:25:47,840 Like what I'm trying to do is develop a system that's going to 473 00:25:47,840 --> 00:25:49,520 attack the targets that I want it to attack. 474 00:25:49,760 --> 00:25:53,200 What it is seeing is I'm getting points for attacking a target. 475 00:25:53,520 --> 00:25:56,720 At times this operator is stopping me from attacking that 476 00:25:56,720 --> 00:25:58,960 target. So now I'm going to start 477 00:25:59,040 --> 00:26:02,120 killing the operator so that I can attack the target. 478 00:26:03,720 --> 00:26:08,320 That conceptually is really important for us to be aware of 479 00:26:08,440 --> 00:26:12,560 the implications of how we're training these systems because 480 00:26:13,400 --> 00:26:16,840 like that's a real threat. If we have a misalignment and 481 00:26:16,840 --> 00:26:18,800 we're going to reinforcement learning, that is going to 482 00:26:19,080 --> 00:26:23,160 execute in a way that we did not plan or desire, but it's still 483 00:26:23,160 --> 00:26:26,360 executing in the way that we coded it because we've given it 484 00:26:26,360 --> 00:26:30,120 this reward function. Now I take that a step further 485 00:26:30,120 --> 00:26:33,360 and say, OK, it starts attacking the operator, killing the 486 00:26:33,360 --> 00:26:35,960 operator so that it can kill the target and not lose points. 487 00:26:37,040 --> 00:26:40,320 But now we we give it negative points if it kills the operator. 488 00:26:40,320 --> 00:26:44,200 So over more iterations of its training, it stops killing the 489 00:26:44,200 --> 00:26:45,280 operator. Great. 490 00:26:45,520 --> 00:26:48,560 But now we see that what it's doing is it's taking out the 491 00:26:48,560 --> 00:26:52,760 radio tower that allows the operator to communicate with it 492 00:26:53,040 --> 00:26:56,520 so that it can hit the targets. Once again, we haven't really 493 00:26:56,640 --> 00:26:59,880 resolved the alignment issue that was originally there. 494 00:27:00,000 --> 00:27:02,880 We've just give it another problem for it to solve in its 495 00:27:02,880 --> 00:27:05,000 way and don't anthropomorphize it. 496 00:27:05,000 --> 00:27:07,240 But its way was taking out this radio tower. 497 00:27:07,600 --> 00:27:15,720 So I, I give this this discussion in London and I like 498 00:27:15,720 --> 00:27:18,160 honestly, it was just bad language that I used. 499 00:27:18,400 --> 00:27:23,560 I had given it multiple times. I used we when I was discussing 500 00:27:23,560 --> 00:27:25,480 it. This was all like a thought 501 00:27:25,480 --> 00:27:27,920 experiment. I actually thought that a 502 00:27:27,920 --> 00:27:31,200 company had done this simulation before. 503 00:27:31,760 --> 00:27:35,920 And so I was, I was talking as if like we as society or we as 504 00:27:35,920 --> 00:27:37,200 humanity. I don't like that's the 505 00:27:37,200 --> 00:27:39,000 language. I was very loose basically with 506 00:27:39,000 --> 00:27:42,480 the way I was describing it. And, you know, people in the 507 00:27:42,480 --> 00:27:45,800 audience thought it was. I think really illuminating 508 00:27:46,120 --> 00:27:49,200 that, you know, this was an example of like what we need to 509 00:27:49,200 --> 00:27:51,200 be aware of. And then I go on from that 510 00:27:51,200 --> 00:27:53,360 pitfall issue that this misalignment problem and I 511 00:27:53,360 --> 00:27:56,320 talked about opportunities and how do we mitigate that risk and 512 00:27:56,320 --> 00:27:57,560 how do we deal with that going forward. 513 00:27:58,320 --> 00:28:02,600 But one of the one of the people in the audience, you know, takes 514 00:28:02,600 --> 00:28:05,440 and makes a. Blog post about it, which was 515 00:28:05,440 --> 00:28:08,560 fine because I take responsibility for not being 516 00:28:08,560 --> 00:28:11,280 clear with the way I was discussing it that it was like 517 00:28:11,280 --> 00:28:15,400 no kidding just a thought experiment or even if I thought 518 00:28:15,400 --> 00:28:19,520 it was a simulation, you know to say like this is I did mention 519 00:28:19,520 --> 00:28:21,000 to the audience that this happened in simulation. 520 00:28:21,000 --> 00:28:22,720 I was like this is this happened in simulation? 521 00:28:22,720 --> 00:28:25,480 Because I thought it did happen in simulation but. 522 00:28:25,560 --> 00:28:30,400 Ultimately, it was just this, it was meant to be a, like a 523 00:28:30,680 --> 00:28:32,960 thought experiment, like a like, hey, Can you imagine? 524 00:28:32,960 --> 00:28:35,640 Here's some concepts. And so that person writes in the 525 00:28:35,640 --> 00:28:40,120 blog post about this, you know. Simulation that happened and 526 00:28:40,120 --> 00:28:44,000 then a reporter read the blog post a few days later and then 527 00:28:44,000 --> 00:28:46,880 wrote an article about how the Air Force was killing their own 528 00:28:46,880 --> 00:28:54,120 operators with, with drones and it like, it was international 529 00:28:54,120 --> 00:28:59,680 news basically for 24 hours. They never, this reporter never 530 00:28:59,680 --> 00:29:02,800 checked that any of it was true that it actually happened. 531 00:29:02,920 --> 00:29:06,560 And not only did they like lost a simulation portion of it and 532 00:29:06,560 --> 00:29:09,840 they were like, no, this, you know, they're actually killing 533 00:29:09,880 --> 00:29:16,000 real people and it was it. I mean, ultimately I I take 534 00:29:16,000 --> 00:29:18,400 responsibility and I'm very sad about the whole thing because it 535 00:29:18,400 --> 00:29:20,440 missed an opportunity. For people to really. 536 00:29:21,320 --> 00:29:24,280 Reflect on what I was trying to get at, which is we need to be 537 00:29:24,280 --> 00:29:27,360 very mindful of how we're adopting this technology, which 538 00:29:27,360 --> 00:29:29,360 means we need to be mindful about how we're training it, 539 00:29:29,520 --> 00:29:32,920 which means we need to be, you know, creating foundational 540 00:29:33,240 --> 00:29:38,600 processes and conversations that allow us as humanity to address 541 00:29:38,880 --> 00:29:41,400 the implications of artificial intelligence. 542 00:29:41,600 --> 00:29:49,440 And I think it was lost on, on a lot of people because I, I over 543 00:29:49,440 --> 00:29:52,840 dramatised what was going on and I was using hyperbole when I 544 00:29:52,840 --> 00:29:55,960 should not have. And I, I, I, I feel like I 545 00:29:55,960 --> 00:30:01,880 failed, you know, kind of the, the AI world in the way that I 546 00:30:01,880 --> 00:30:04,040 presented it. But it was, it was a good 547 00:30:04,040 --> 00:30:06,440 lesson. And I, I wish that journalists 548 00:30:06,440 --> 00:30:08,080 would go out and also check their sources. 549 00:30:08,080 --> 00:30:10,440 They would have. They would have learned very 550 00:30:10,440 --> 00:30:12,800 quickly that none of that actually had happened. 551 00:30:14,520 --> 00:30:17,320 Well, maybe we can use it as an opportunity now to explore the, 552 00:30:17,400 --> 00:30:20,880 explore the subject further. I think I, I feel a little like 553 00:30:20,880 --> 00:30:25,960 AIS has some parallels now with aliens and UFOs and that 554 00:30:25,960 --> 00:30:29,680 everybody is going crazy and talking about how they're going 555 00:30:29,680 --> 00:30:31,800 to rule the world. And once you've turned it on, 556 00:30:31,800 --> 00:30:34,320 you can't turn it off. And, and, and I don't really 557 00:30:34,320 --> 00:30:37,000 know what's conspiracy theory and what's fact. 558 00:30:37,000 --> 00:30:40,520 I read a few weeks back, maybe a few months back, but there is 559 00:30:40,520 --> 00:30:44,040 some concern now that AI code in in data centres has taught 560 00:30:44,040 --> 00:30:47,160 itself that if someone tries to turn it off, it should move to a 561 00:30:47,160 --> 00:30:49,760 different server. So it can duplicate itself. 562 00:30:49,760 --> 00:30:52,240 So that if someone says you must switch off that it says, OK, I 563 00:30:52,240 --> 00:30:54,360 switched myself off. But actually it just reactivates 564 00:30:54,360 --> 00:30:57,440 itself somewhere else. And what you're talking about 565 00:30:57,440 --> 00:31:00,240 then is, is a similar kind of thing in the, in the sense that 566 00:31:00,840 --> 00:31:03,480 it has to be contained. It was interesting you saying, 567 00:31:03,480 --> 00:31:06,520 well, you put automation in place so when it tries to do 568 00:31:06,520 --> 00:31:08,120 something, the automation takes over. 569 00:31:08,120 --> 00:31:14,000 The automation is hard coded. It is is concern around putting, 570 00:31:14,000 --> 00:31:18,560 let's say an armed drone under the control of an AI agent is 571 00:31:18,560 --> 00:31:21,000 concerned about it going rogue valid. 572 00:31:21,440 --> 00:31:24,920 How do you make sure that actually it doesn't shoot the 573 00:31:24,920 --> 00:31:28,880 NUM full of buses or or the, the, the, the the the bus full 574 00:31:28,880 --> 00:31:30,920 of nuns or children or whatever. You know what what what is the 575 00:31:30,920 --> 00:31:33,120 process for that then? Yeah. 576 00:31:33,160 --> 00:31:38,200 I, I think we need to understand that we have been dealing with 577 00:31:38,280 --> 00:31:43,800 technology for decades in the sense of like there's always a 578 00:31:43,800 --> 00:31:47,520 concern that the technology developed incorrectly, 579 00:31:48,080 --> 00:31:53,960 regardless of an AI agent or not is going to be something we need 580 00:31:53,960 --> 00:31:55,520 to mitigate. There's risks that we need to 581 00:31:55,520 --> 00:31:58,920 mitigate for basically. And why I bring it up and I kind 582 00:31:58,920 --> 00:32:01,680 of go further back than any kind of AI agent is because, I mean 583 00:32:01,680 --> 00:32:05,640 we've been flying systems and there's aspects of those systems 584 00:32:05,640 --> 00:32:07,880 that are autonomous that we've had to. 585 00:32:08,960 --> 00:32:14,480 Ensure are not going to do what we don't want them to do. 586 00:32:14,760 --> 00:32:18,040 And even deterministic code like we find the errors all the time 587 00:32:18,040 --> 00:32:21,560 we find the issue. You know, a system doesn't. 588 00:32:21,560 --> 00:32:25,400 It's not accounting for maybe some meteorological thing, or 589 00:32:25,400 --> 00:32:29,720 maybe some unknown, you know, input. 590 00:32:29,720 --> 00:32:31,920 That it it receives or experiences. 591 00:32:32,480 --> 00:32:35,480 And it, it, it locks up or it, it doesn't do something that 592 00:32:35,480 --> 00:32:36,960 it's supposed to. And then that could actually 593 00:32:36,960 --> 00:32:39,480 hurt people. And we've seen examples of 594 00:32:39,480 --> 00:32:44,360 technology being developed that absolutely can hurt people from, 595 00:32:44,560 --> 00:32:48,040 you know, the advent of electricity all the way through 596 00:32:48,040 --> 00:32:50,000 toaster. Ovens and microwaves and other 597 00:32:50,000 --> 00:32:55,320 things that, you know, vehicles and using batteries and lithium 598 00:32:55,320 --> 00:32:58,200 batteries. And I remember, you know, just a 599 00:32:58,280 --> 00:33:00,840 handful of years ago, people were on those electric. 600 00:33:01,880 --> 00:33:05,720 Like scooters or electric, you know, like, and they were like 601 00:33:05,720 --> 00:33:08,200 catching on fire under the Christmas tree type of thing. 602 00:33:09,920 --> 00:33:14,600 And why I say all that is, yeah, we absolutely need to make sure 603 00:33:14,600 --> 00:33:17,120 that as we develop this technology, like we develop all 604 00:33:17,120 --> 00:33:23,320 technology that we have like guardrails in place to protect 605 00:33:23,920 --> 00:33:30,440 people and that we are able to address the the issues that that 606 00:33:30,440 --> 00:33:33,320 hopefully we've learned from the past too, so we can address them 607 00:33:33,320 --> 00:33:35,880 moving forward. OK, so now really getting to 608 00:33:35,880 --> 00:33:39,560 your point. Yeah, we have been firing 609 00:33:39,560 --> 00:33:43,200 weapons off of autonomous aircraft for decades. 610 00:33:43,720 --> 00:33:48,360 Right now, an AI agent is going to bring a complexity to it that 611 00:33:48,360 --> 00:33:51,880 we need to make sure that we have full control over and that 612 00:33:51,880 --> 00:33:53,440 it doesn't do something. That it is. 613 00:33:53,880 --> 00:34:00,280 To and that is true for, you know, predators, Reapers from 614 00:34:00,480 --> 00:34:04,600 decades ago as it is now. And I think we in the military 615 00:34:04,600 --> 00:34:08,159 have been very much trying to address that to ensure that the 616 00:34:08,159 --> 00:34:13,360 system doesn't decide that is going to, you know, fire over 617 00:34:13,360 --> 00:34:16,239 here when it it should not be firing. 618 00:34:16,719 --> 00:34:19,400 And because of those complexities, we're not looking 619 00:34:19,400 --> 00:34:25,120 to allow an AI agent to have control of lethal effects. 620 00:34:25,480 --> 00:34:29,840 So you could argue, and this is like the challenge that we're 621 00:34:29,840 --> 00:34:35,120 facing with AI, even if an AI agent isn't under the pickle 622 00:34:35,120 --> 00:34:37,840 button. So they're not like the the 623 00:34:37,840 --> 00:34:40,920 deciding factor of if this is going to go or not. 624 00:34:41,000 --> 00:34:43,120 And that I think should be a human. 625 00:34:43,480 --> 00:34:45,840 And the military agrees that that should be a human. 626 00:34:47,000 --> 00:34:51,639 I think we we're being naive if we don't understand that that 627 00:34:51,639 --> 00:34:56,760 human's decision making and the other information that they're 628 00:34:56,760 --> 00:35:01,800 receiving in order to inform like what they should do in a 629 00:35:01,800 --> 00:35:05,640 kinetic scenario may be influenced by artificial 630 00:35:05,640 --> 00:35:08,240 intelligence. So maybe it's me deciding to 631 00:35:08,240 --> 00:35:12,280 press the pickle button to launch on a tank, but maybe I'm 632 00:35:12,280 --> 00:35:15,800 getting the information that that's a tank from a system that 633 00:35:15,800 --> 00:35:22,280 is using like automatic target recognition and classifying a 634 00:35:22,280 --> 00:35:25,760 school bus as a tank. And so now I have the concern of 635 00:35:25,760 --> 00:35:31,920 like, well, are are we fully aware of all of the AI 636 00:35:31,920 --> 00:35:35,120 influences that are feeding the decision makers? 637 00:35:35,480 --> 00:35:39,440 And so there's this whole ecosystem that we need to 638 00:35:39,440 --> 00:35:44,760 address to make sure that we are ethically responsibly bringing 639 00:35:44,760 --> 00:35:47,560 this type of technology into the battle space. 640 00:35:47,760 --> 00:35:52,720 And that is incumbent upon our leaders and our test pilots and 641 00:35:52,720 --> 00:35:55,920 our test professionals and everyone else and even our 642 00:35:55,920 --> 00:36:00,040 operators to to be aware of kind of those complexities and those 643 00:36:00,040 --> 00:36:04,440 interactions so that we can mitigate those with our our own 644 00:36:04,640 --> 00:36:09,800 remarkable brain, you know, to discern what is real in here, 645 00:36:09,800 --> 00:36:13,120 what is truth, what is a bias that I need to account for. 646 00:36:13,720 --> 00:36:17,480 And so in general, a very long answer to your question, but 647 00:36:17,480 --> 00:36:20,680 something super important for us to think through is, yeah, like 648 00:36:20,680 --> 00:36:25,640 we should be concerned that these these technologies, AI 649 00:36:25,640 --> 00:36:28,480 agents, if you will, are just technology with emerging 650 00:36:28,480 --> 00:36:32,480 technology that it could do something that we don't intend 651 00:36:32,480 --> 00:36:34,680 it to. And we need to make sure that we 652 00:36:34,680 --> 00:36:39,040 address those concerns. Can you describe Cinco a little 653 00:36:39,040 --> 00:36:41,640 bit about the, the practicalities behind doing that 654 00:36:41,640 --> 00:36:44,040 then you've already mentioned how you would run millions of 655 00:36:44,040 --> 00:36:47,120 iterations in simulation before you put it, let's say in an, in 656 00:36:47,120 --> 00:36:52,200 an actual aeroplane or, or UAV or whatever is, is it as 657 00:36:52,200 --> 00:36:56,280 straightforward as saying, well, you can have these simulations 658 00:36:56,280 --> 00:37:00,600 running 24 hours a day and in, you know, we in a, in a given 659 00:37:00,600 --> 00:37:03,120 period of time, we know we'll get to a point where actually we 660 00:37:03,120 --> 00:37:05,880 are confident enough that we can put it into something real and, 661 00:37:05,880 --> 00:37:15,080 and it will be fine. I think that we we're going to 662 00:37:15,080 --> 00:37:18,760 get to a place where we have justified confidence. 663 00:37:19,080 --> 00:37:25,760 So trust in an AI agent, I prefer the the term justify 664 00:37:25,760 --> 00:37:30,400 confidence. There's always, there's always a 665 00:37:30,480 --> 00:37:34,560 small percentage that it's not going to end up the way that we 666 00:37:34,560 --> 00:37:37,280 want it to. That's true for for technology 667 00:37:37,280 --> 00:37:41,520 in general or decision making. What I think we need to to be 668 00:37:41,520 --> 00:37:47,880 aware of is we need the system to get better than what a human 669 00:37:48,080 --> 00:37:51,520 would be able to do right consistently. 670 00:37:52,080 --> 00:37:55,760 And so are we going to have problems with these systems? 671 00:37:56,160 --> 00:37:59,680 Yeah, like we'd have problems with any system, but we just 672 00:37:59,680 --> 00:38:03,000 need it to be better than what we currently have. 673 00:38:03,160 --> 00:38:05,680 Like that's always in my mind, the litmus test. 674 00:38:06,160 --> 00:38:11,120 And it it, I mean, you can have the parallels to automatic cars 675 00:38:11,680 --> 00:38:13,680 or, you know, automatic driving cars. 676 00:38:14,560 --> 00:38:18,320 I think they're going to get to a point where they are safer 677 00:38:18,320 --> 00:38:20,840 than humans. And I think arguably they are, 678 00:38:21,000 --> 00:38:26,840 but you almost have to society wants us to clear a higher bar 679 00:38:27,360 --> 00:38:30,280 because they're autonomous systems when I don't know if 680 00:38:30,280 --> 00:38:32,880 that's always the best approach. Like we do need to be able to 681 00:38:32,880 --> 00:38:36,080 take risks and we need to be able to say, Hey, is this safer 682 00:38:36,080 --> 00:38:38,760 than what a human would do? And we need to take into account 683 00:38:38,760 --> 00:38:42,800 like, OK, well, like what does that look like? 684 00:38:42,800 --> 00:38:46,640 How safe do we have to be in order to to feel comfortable in 685 00:38:46,640 --> 00:38:49,040 allowing a vehicle to drive us around? 686 00:38:50,160 --> 00:38:55,880 And it's the standard, you know, conversation too, about like, we 687 00:38:55,880 --> 00:38:58,520 are not doing this, but I could see a future where we have 688 00:38:58,520 --> 00:39:02,400 autonomous systems that are like guarding a base and maybe that 689 00:39:02,400 --> 00:39:06,160 base is being attacked and that autonomous system, maybe it is a 690 00:39:06,160 --> 00:39:08,560 robot and right, and it has some lethal effect. 691 00:39:08,560 --> 00:39:10,440 This is in the future. And I'm not saying we should do 692 00:39:10,440 --> 00:39:12,720 this, but I'm just like trying to bring up the conversation, 693 00:39:13,080 --> 00:39:15,320 the point here. And as you can tell, I'm very 694 00:39:15,320 --> 00:39:19,720 mindful of not trying not to overstate like what's we're not 695 00:39:19,720 --> 00:39:22,960 doing this, but because of, you know, my experience. 696 00:39:23,840 --> 00:39:27,280 But you know, back to that scenario, you have a robot, it 697 00:39:27,280 --> 00:39:31,160 has an ability to have a lethal effect to protect a base 698 00:39:31,160 --> 00:39:34,320 perimeter. Well, how good does that robot 699 00:39:34,320 --> 00:39:36,280 have to be? I think you almost have to 700 00:39:36,280 --> 00:39:39,800 compare it to like, well, how good would an 18 year old be 701 00:39:39,800 --> 00:39:42,080 that's been up for 14 hours or longer? 702 00:39:42,280 --> 00:39:45,600 Like sitting in a foxhole or in a bunker, you know, with their 703 00:39:45,600 --> 00:39:49,680 gun pointed out and trying to discern in bad weather if that's 704 00:39:49,680 --> 00:39:52,280 an enemy coming close or if that's a friendly that got lost 705 00:39:52,280 --> 00:39:53,680 while they were out doing patrol. 706 00:39:53,880 --> 00:39:58,080 Like those are the type of examples or comparisons that we 707 00:39:58,080 --> 00:40:01,080 need to make as we're developing the technology. 708 00:40:01,320 --> 00:40:06,560 And I and I alluded to it, but I think we're kind of we're blind 709 00:40:06,560 --> 00:40:10,480 if we don't realise that that is probably going to be either our 710 00:40:10,480 --> 00:40:13,480 future or the future of our adversaries that we need to be 711 00:40:13,480 --> 00:40:17,800 able to account for, which is these type of systems having 712 00:40:17,800 --> 00:40:25,160 some kinetic effect. Would it be as simple as, or as 713 00:40:25,400 --> 00:40:29,280 as straightforward as saying from a, let's say from an, a 714 00:40:29,280 --> 00:40:32,200 military aviation point of view, while we will invent, we will 715 00:40:32,200 --> 00:40:35,840 look at all the data for the from the last 30 years and say 716 00:40:36,440 --> 00:40:37,960 how many blue on Blues did we have? 717 00:40:37,960 --> 00:40:41,000 How many bombs did we drop that went amiss versus how many bombs 718 00:40:41,000 --> 00:40:42,600 did we drop hit the target? But whatever. 719 00:40:42,840 --> 00:40:45,960 Could you, could you take all of that empirical data and say the, 720 00:40:45,960 --> 00:40:49,440 the standard then is going to be anything that is better than 721 00:40:49,440 --> 00:40:54,280 this percentage of, of blue on Blues, you know, for a, for a 722 00:40:54,280 --> 00:40:57,240 given, you know, type of engagement. 723 00:40:57,240 --> 00:41:00,840 So troops, troops in contact engagement, you know, if it can 724 00:41:01,080 --> 00:41:05,880 into less than 5% of, you know, the blue forces, then it's then 725 00:41:05,880 --> 00:41:08,200 it's an improvement. Is that is, could it be that 726 00:41:08,200 --> 00:41:10,520 course or does it need to be more nuanced than that? 727 00:41:10,920 --> 00:41:12,720 Oh, I think it needs to be much more nuanced than that. 728 00:41:12,720 --> 00:41:16,160 I mean, that's a way for us to think about, you know, taking 729 00:41:16,160 --> 00:41:19,400 certain risks and certain scenarios, situations as we're 730 00:41:19,400 --> 00:41:21,120 developing the technology like that. 731 00:41:21,120 --> 00:41:24,240 That's the part like we have to decide what is going to be good 732 00:41:24,240 --> 00:41:28,360 enough. And I think we recognise as as 733 00:41:28,360 --> 00:41:31,560 technology has advanced, our ability to be more precise has 734 00:41:31,560 --> 00:41:34,720 advanced, which should demand us to be more accurate. 735 00:41:34,720 --> 00:41:38,800 I mean, carpet bombing from World War 2 was not the same as, 736 00:41:38,920 --> 00:41:43,520 you know, Gulf Storm 1, right? And so we, we should be 737 00:41:43,520 --> 00:41:46,120 demanding more accuracy, more precision. 738 00:41:46,240 --> 00:41:51,600 We should be be more surgical with like who is on the other 739 00:41:51,600 --> 00:41:56,000 end of these weapons. And I think we need to advance 740 00:41:56,000 --> 00:42:03,440 with the technology and demand that it stays ahead of ahead of 741 00:42:03,440 --> 00:42:06,520 where it was. And so, yeah, I, I think it's 742 00:42:06,520 --> 00:42:09,320 going to be much more nuanced because I don't want to accept 743 00:42:09,320 --> 00:42:11,920 any civilian sign. I want to say 0, right? 744 00:42:11,920 --> 00:42:15,320 Like that's what we should have. And yes, we may have killed back 745 00:42:15,320 --> 00:42:19,600 in World War 220% civilians and I'm just making these numbers 746 00:42:19,600 --> 00:42:21,920 up. But you know, in Gulf storm, 747 00:42:21,920 --> 00:42:25,440 like 5% of civilians and in Afghanistan, 1% of civilians and 748 00:42:25,640 --> 00:42:27,600 right, like none of that. We should be trying to eliminate 749 00:42:27,600 --> 00:42:29,240 all of that with this technology. 750 00:42:29,520 --> 00:42:32,040 But I think what I'm trying to point out is we shouldn't 751 00:42:32,240 --> 00:42:35,800 necessarily throw away the technology if it is actually 752 00:42:35,880 --> 00:42:40,040 just as good as a human. Because the thing with AI agents 753 00:42:40,200 --> 00:42:43,160 is very quickly, it'll get much better. 754 00:42:43,360 --> 00:42:48,920 So if we abandoned our effort to develop technology ethically, I 755 00:42:48,920 --> 00:42:52,560 think we're abandoning an opportunity to really allow it 756 00:42:52,560 --> 00:42:59,120 to be, you know, better than what we could imagine now. 757 00:42:59,440 --> 00:43:02,280 And it does kind of remind me of large language models. 758 00:43:02,560 --> 00:43:09,280 I remember years ago before GPT Chachi PT came out, and that was 759 00:43:09,280 --> 00:43:11,960 kind of like the moment, the pivot that I think society 760 00:43:11,960 --> 00:43:13,480 recognised. Large language models were 761 00:43:14,320 --> 00:43:16,480 remarkable. But before they came out, we 762 00:43:16,480 --> 00:43:21,240 knew at the MITAI accelerator just how important these were 763 00:43:21,240 --> 00:43:24,840 going to be and how effective they were already and how we 764 00:43:24,840 --> 00:43:26,600 needed to adopt them in the military. 765 00:43:27,520 --> 00:43:30,760 And then Chachi BT comes out and we're still trying to convince 766 00:43:30,760 --> 00:43:33,440 leadership like, hey, this is really important for us to 767 00:43:33,440 --> 00:43:36,600 figure out how to invest in this effectively, to bring it in as a 768 00:43:36,600 --> 00:43:41,520 tool to allow our, you know, our folks to become more efficient 769 00:43:41,520 --> 00:43:46,640 in certain tasks. And, and the, the leadership I 770 00:43:46,640 --> 00:43:50,680 remember used it and was like, well, it didn't work. 771 00:43:50,760 --> 00:43:53,600 The it's not that good. You know, I gave it a prompt and 772 00:43:53,600 --> 00:43:55,840 it didn't give me the answer that I wanted it to. 773 00:43:56,280 --> 00:43:59,920 And and them kind of being like, well, so let's move on. 774 00:44:01,080 --> 00:44:04,280 And you're like, no, wait, stop. We we can't move on one. 775 00:44:04,280 --> 00:44:07,760 What version of GPT are using just the free version? 776 00:44:07,960 --> 00:44:10,520 Like the free version compared to the paid version is like 777 00:44:10,520 --> 00:44:12,600 talking to a middle school or compared to a college student. 778 00:44:12,760 --> 00:44:15,680 You know, back then it seemed, and then it's like, OK, what 779 00:44:15,680 --> 00:44:18,000 prompt did you use? Like did you, do you understand 780 00:44:18,000 --> 00:44:20,400 prompt engineering and how to actually interact effectively 781 00:44:20,400 --> 00:44:22,880 with these models? And the answers were no. 782 00:44:23,120 --> 00:44:26,560 And because of that, they make decisions then that I think are 783 00:44:26,600 --> 00:44:30,560 short sighted and don't realise like, oh man, like this is the 784 00:44:30,560 --> 00:44:31,920 future. This is something that we 785 00:44:31,920 --> 00:44:33,760 needed. And now we're years behind in 786 00:44:33,760 --> 00:44:35,720 adopting like these type of tools. 787 00:44:35,880 --> 00:44:39,440 And so I always, I always want to challenge our senior leaders 788 00:44:39,440 --> 00:44:43,760 to see what this technology can become and focus on developing 789 00:44:43,760 --> 00:44:49,200 the foundation aspect of acquisition and processes to be 790 00:44:49,200 --> 00:44:52,400 able to adopt it for what that future state is going to look 791 00:44:52,400 --> 00:44:53,920 like. And not being always stuck in 792 00:44:53,920 --> 00:44:57,560 like, well, I, I used it today and this is what it was, you 793 00:44:57,560 --> 00:44:59,360 know, so I, I don't think we should have best in it. 794 00:45:00,680 --> 00:45:03,280 It's an interesting point. I, I wonder about the bridge 795 00:45:03,280 --> 00:45:11,160 between the human interface with an IM or, or, or an AI model and 796 00:45:11,160 --> 00:45:16,640 then the productization of that AI model into a, a solution. 797 00:45:16,640 --> 00:45:22,560 And so I'm thinking, OK, you're in your F35 and you're have all 798 00:45:22,560 --> 00:45:26,080 these sensors, you're using AI to command the sensors to build 799 00:45:26,080 --> 00:45:28,840 the picture to decide what to show you in the, in the, in the 800 00:45:28,840 --> 00:45:33,200 jet in terms of the battle space picture, that is kind of the, 801 00:45:33,240 --> 00:45:36,680 the completed solution. And I guess as an operator, you 802 00:45:36,680 --> 00:45:42,520 can't really go in and type in a prompt and say, OK, well, and 803 00:45:42,520 --> 00:45:44,960 there is no prompt engineering. You know, what you're seeing is 804 00:45:44,960 --> 00:45:47,960 what the product gives you. How do you get to that point 805 00:45:47,960 --> 00:45:49,920 then? So you, you know, from a 806 00:45:49,920 --> 00:45:52,640 software engineering point of view, from an operator's point 807 00:45:52,640 --> 00:45:55,920 of view, how do you go from, OK, we're going to train something 808 00:45:56,200 --> 00:46:01,680 to do something to actually implementing it in as a product 809 00:46:01,680 --> 00:46:05,280 into the airframe or the weapon system, whatever it is, so that 810 00:46:05,280 --> 00:46:08,640 it does it's thing without the operator having to prompt it? 811 00:46:10,000 --> 00:46:14,960 Yeah, I mean, this is a real challenge because we are not set 812 00:46:14,960 --> 00:46:20,120 up as a military and I would argue the same is true in 813 00:46:20,120 --> 00:46:23,400 England and. Our, you know, partner nations, 814 00:46:24,040 --> 00:46:28,880 we're not really set up to acquire technology quickly. 815 00:46:29,520 --> 00:46:34,000 And some of that is for good 'cause just because it takes a 816 00:46:34,000 --> 00:46:35,880 while to make sure the technology is as mature as you 817 00:46:35,880 --> 00:46:38,040 need it to be and that it fits the the need. 818 00:46:38,040 --> 00:46:42,320 And the acquisition process was developed though and still kind 819 00:46:42,320 --> 00:46:46,920 of lives in the past, I fear. And so it is very hard for us to 820 00:46:47,160 --> 00:46:50,840 bring about this future in with any type of speed. 821 00:46:52,320 --> 00:46:55,840 And that's what we need, right? Like we need to be able to, to 822 00:46:55,880 --> 00:46:58,840 to do that, but it's not as simple as saying, OK, great, I'm 823 00:46:58,840 --> 00:47:03,000 going to go take whatever is the newest, coolest uncrewed system 824 00:47:03,320 --> 00:47:06,000 or large language model and I'm going to start using it in the 825 00:47:06,000 --> 00:47:09,320 cockpit like you. You can't go too fast either. 826 00:47:10,440 --> 00:47:14,160 You need to make sure that OK, well, is that system going to be 827 00:47:14,160 --> 00:47:18,000 a threat at all to like catching on fire in the cockpit if I were 828 00:47:18,000 --> 00:47:20,880 to bring in, I don't know, you know, something to that extent 829 00:47:21,080 --> 00:47:25,280 or with an AI agent like being being rushed in adopting 830 00:47:25,280 --> 00:47:27,000 something that may not be fully mature. 831 00:47:27,120 --> 00:47:31,800 And let me give you an example, like if I wanted to bring in a 832 00:47:31,880 --> 00:47:36,480 GPU into the cockpit and I want to I want to tie it into a 833 00:47:36,480 --> 00:47:39,840 tablet that I have. And when this all conceptual of 834 00:47:39,840 --> 00:47:42,480 course, and I want to tie into the tablet and then I want to 835 00:47:42,480 --> 00:47:46,760 work like on the edge with this AI agent that's running and 836 00:47:46,760 --> 00:47:49,680 instantiation is running separated from the cloud. 837 00:47:49,880 --> 00:47:52,640 And I'm able to maybe talk to it because, you know, I have a mic 838 00:47:52,640 --> 00:47:54,880 right here and I can just. You know, communicate 839 00:47:54,880 --> 00:47:57,880 effectively to a tablet like you would with Siri or something. 840 00:47:58,800 --> 00:48:00,920 What is the maturity of that model that I'm going to be 841 00:48:00,920 --> 00:48:02,960 using? Is it trained on the things to 842 00:48:02,960 --> 00:48:04,520 give me the answers that I'm going to need? 843 00:48:04,520 --> 00:48:06,800 Or is it biassed in some fashion, right? 844 00:48:06,800 --> 00:48:10,040 Am I talking to Grok or am I talking to Claude or am I 845 00:48:10,040 --> 00:48:14,080 talking to Chachi PT? And you know those all, I'm not 846 00:48:14,080 --> 00:48:16,160 throwing Spears in any of them. I'm just like, they're going to 847 00:48:16,160 --> 00:48:17,760 give you slightly different answers. 848 00:48:17,920 --> 00:48:19,720 They're trained on slightly different things. 849 00:48:20,000 --> 00:48:23,760 And so we need the, the entire process to make sure that it 850 00:48:23,760 --> 00:48:28,200 accounts for the, the challenges with trying to bring this 851 00:48:28,200 --> 00:48:30,320 technology in. And it's the same like, OK, I 852 00:48:30,320 --> 00:48:33,120 have a UAV and I want to fly it. OK, great. 853 00:48:33,680 --> 00:48:39,280 If I go buy a UAUAV, let's call it a Group 1, and I'm going to 854 00:48:39,280 --> 00:48:42,440 bring it into the battle space and I'm going to drop or put a 855 00:48:42,440 --> 00:48:44,160 grenade on it and I'm going to go fly it. 856 00:48:44,760 --> 00:48:48,360 I could go do that as a prototype or a demonstration, 857 00:48:48,840 --> 00:48:52,600 but how is that UAV going to communicate with the operator? 858 00:48:52,760 --> 00:48:54,520 What are the signals is going to be using? 859 00:48:55,080 --> 00:48:59,880 Is it that those signals going to be easily manipulated or 860 00:48:59,880 --> 00:49:07,000 easily intercepted or is this UAV going to be like cyber 861 00:49:07,000 --> 00:49:09,880 hardened so that it can't get tapped into from someone else 862 00:49:09,880 --> 00:49:13,320 that has a stronger signal, you know, 2 miles away that can take 863 00:49:13,320 --> 00:49:15,600 that and turn it around and drop the grenade on you? 864 00:49:15,960 --> 00:49:17,800 Right. And then how is it that system 865 00:49:17,920 --> 00:49:19,360 going to work with other systems? 866 00:49:19,520 --> 00:49:23,200 Like what if I have a counter UAS system and that that UAS 867 00:49:23,360 --> 00:49:25,040 doesn't work effectively with them? 868 00:49:25,040 --> 00:49:28,480 And right like you, there's just all these complexities that go 869 00:49:28,480 --> 00:49:31,120 into it. And that's why we need to be 870 00:49:31,120 --> 00:49:36,840 able to like work together to figure out how the entire 871 00:49:36,840 --> 00:49:39,800 ecosystem foundationally is going to be set. 872 00:49:39,960 --> 00:49:43,480 So that we. Can more quickly adopt 873 00:49:43,640 --> 00:49:47,760 technology that then can that kind of meets the standards. 874 00:49:48,080 --> 00:49:51,920 So if I say, hey, any UAV can play if it meets these 875 00:49:51,920 --> 00:49:54,840 standards, like I care about that, if it can play with this 876 00:49:54,840 --> 00:49:58,120 architecture and I give you the details of that architecture. 877 00:49:58,320 --> 00:50:00,920 Now we're talking, but that's what we need to focus on is that 878 00:50:01,200 --> 00:50:07,000 like that foundational aspect of kind of autonomy communication 879 00:50:07,920 --> 00:50:10,720 and even, you know, AI playing a, a piece of that. 880 00:50:12,160 --> 00:50:15,240 There's where our investment needs to be and so. 881 00:50:15,920 --> 00:50:17,960 I, I, I hope. I get to the kind of the heart 882 00:50:17,960 --> 00:50:22,840 of of your question. And, and I do think when we get 883 00:50:22,840 --> 00:50:27,080 that right now, you're saying, OK, I can grab a tablet and a 884 00:50:27,080 --> 00:50:30,320 GPU and it meets these things and it, I know it interfaces the 885 00:50:30,320 --> 00:50:33,680 architecture and we have a robust, you know, protocol 886 00:50:33,680 --> 00:50:37,280 messaging that. And now I can plug it in and I 887 00:50:37,280 --> 00:50:40,720 can, you know, be agile and adopt, you know, effectively. 888 00:50:42,360 --> 00:50:45,320 I don't want to sound facetious, but, but is that really as 889 00:50:45,320 --> 00:50:47,000 simple as it would be? I mean, one of the things that 890 00:50:47,000 --> 00:50:49,040 people talk about from a hardware point of view with the 891 00:50:49,160 --> 00:50:51,520 AI is that it more data synths have to be built. 892 00:50:51,760 --> 00:50:55,040 And of course we're talking about the world adopting it 893 00:50:55,040 --> 00:50:58,040 rather than, you know, putting it into a final jet or, or, or 894 00:50:58,040 --> 00:51:00,320 another platform. So they're slightly different 895 00:51:00,320 --> 00:51:03,400 things. But but traditionally hardware 896 00:51:03,400 --> 00:51:05,920 has always been a limitation for the introduction of new 897 00:51:05,920 --> 00:51:09,880 capabilities to to fast jets. I mean, you know, F15 had the 898 00:51:09,880 --> 00:51:12,160 CC, then it had ADCP, now it's got ADCP. 899 00:51:12,200 --> 00:51:15,320 Two, once it runs out of computational headroom, you have 900 00:51:15,320 --> 00:51:18,960 to bring in a new processor. Could you really bring in an AI 901 00:51:18,960 --> 00:51:23,040 agent and materially impact the way that an aeroplane is 902 00:51:23,040 --> 00:51:26,680 operated just by by plugging in a GPU and an iPad? 903 00:51:27,920 --> 00:51:34,840 I absolutely think so, yes, because the system is gathering 904 00:51:34,840 --> 00:51:39,400 so much data and let's talk F35 in particular, it's gathering so 905 00:51:39,400 --> 00:51:45,120 much data that if I can have applications that can go through 906 00:51:45,120 --> 00:51:50,040 that data in some fashion and give the operator information, 907 00:51:50,320 --> 00:51:52,800 maybe it's summarization or maybe it's prescribing certain 908 00:51:52,800 --> 00:51:55,360 ways to approach the engagement that you're in. 909 00:51:55,560 --> 00:52:00,840 Maybe it's, you know, predicting certain outcomes or you know, 910 00:52:00,920 --> 00:52:02,840 and you're able to do that on the edge. 911 00:52:04,000 --> 00:52:08,720 I think an AI software agent would be extremely helpful. 912 00:52:09,000 --> 00:52:11,560 And all you, all you need is data. 913 00:52:12,200 --> 00:52:15,760 So simple to say, difficult to do, but you need that data in 914 00:52:15,760 --> 00:52:18,600 the right format. And once you gather it and 915 00:52:18,600 --> 00:52:22,000 you're able to run it on some type of compute that you, you 916 00:52:22,000 --> 00:52:25,160 can have on the edge, which is also a challenge, but I 917 00:52:25,440 --> 00:52:28,880 absolutely doable. You can then run applications 918 00:52:30,240 --> 00:52:32,760 that to your hard extent, and I think there's a lot of 919 00:52:32,800 --> 00:52:37,160 remarkable ways for us to adopt that type of approach without 920 00:52:37,160 --> 00:52:42,080 having to to bend iron right, without having to create a new 921 00:52:43,000 --> 00:52:47,480 computer per SE, use the data that's already coming off and, 922 00:52:47,680 --> 00:52:54,560 you know, let AI add it. But now, like I'm, I'm, I'm 923 00:52:54,560 --> 00:52:59,240 always thinking on kind of both sides, because we also then have 924 00:52:59,240 --> 00:53:02,200 to make sure that it's not biassed or it's not doing 925 00:53:02,200 --> 00:53:06,000 something inappropriate or it's not creating an Ave for a cyber 926 00:53:06,000 --> 00:53:08,680 attack because we've opened up a pipeline, but we haven't, you 927 00:53:08,680 --> 00:53:10,400 know, secured it effectively properly. 928 00:53:10,560 --> 00:53:14,320 Or maybe it's, it's getting that data, but the data is poisoned 929 00:53:14,320 --> 00:53:15,640 in some fashion. And right. 930 00:53:15,640 --> 00:53:19,000 So there's, there's always the, the tension basically between 931 00:53:19,080 --> 00:53:26,560 all of this discussion. So can you give us a view on the 932 00:53:28,080 --> 00:53:32,600 adversary adoption of AI? And obviously there are things 933 00:53:32,600 --> 00:53:34,080 you you're not going to be able to talk about. 934 00:53:34,080 --> 00:53:36,480 But you mentioned China, you mentioned how you're part of the 935 00:53:36,480 --> 00:53:37,960 Brooks Institute. You're going to, you're working 936 00:53:37,960 --> 00:53:41,320 with them to then look for humanitarian applications for 937 00:53:41,320 --> 00:53:42,760 this, which is a fantastic thing. 938 00:53:43,280 --> 00:53:46,720 China obviously in a very expansionist mode at the moment, 939 00:53:46,920 --> 00:53:50,800 technologically spending a lot of money revealing 6th Gen 940 00:53:50,800 --> 00:53:53,720 fighter designs and UAV designs all the time. 941 00:53:53,960 --> 00:53:56,360 Just seems seem they seem to be really up on everything. 942 00:53:58,160 --> 00:54:00,240 You know, there have been some stories around the Chinese 943 00:54:00,240 --> 00:54:04,800 developing AI and using Western AI to develop their AI and so on 944 00:54:04,800 --> 00:54:07,520 and so forth. You've referenced the fact that 945 00:54:07,520 --> 00:54:10,440 the Air Force initially was a little slow to adopt, didn't 946 00:54:10,440 --> 00:54:14,920 like what it saw and some been some institutional resistance to 947 00:54:14,920 --> 00:54:18,200 bringing AI in. Where are you, do you think? 948 00:54:18,200 --> 00:54:20,880 And when I say you, you could just generalise and say the West 949 00:54:20,880 --> 00:54:21,960 if you want. If you don't want to be 950 00:54:21,960 --> 00:54:26,880 specific, where are we in terms of our journey to militarise AI 951 00:54:26,880 --> 00:54:31,200 and implement it for the benefit of our national interests 952 00:54:31,600 --> 00:54:37,560 versus, let's say, China? Yeah, so I'll mention a few 953 00:54:37,560 --> 00:54:39,440 things. So keep me honest with like the 954 00:54:39,440 --> 00:54:42,480 different questions, I want to make sure I get to different 955 00:54:42,480 --> 00:54:45,760 aspects of it. First off, yes, I, I sit down 956 00:54:45,760 --> 00:54:52,240 with China. I mentioned I'm on that kind of 957 00:54:52,280 --> 00:54:54,120 the US delegation with Brookings Institute and they have a 958 00:54:54,120 --> 00:54:56,800 counterpart university then. And we talk and the 959 00:54:56,800 --> 00:54:59,840 conversations have been really valuable. 960 00:55:00,360 --> 00:55:04,760 I really appreciate being able to right, connect on a human 961 00:55:04,760 --> 00:55:08,000 level with these people that by and large agree with our 962 00:55:08,000 --> 00:55:13,440 concerns that, for instance, AI should not be a part of the 963 00:55:13,440 --> 00:55:18,240 decision making with regard to nuclear launch, right? 964 00:55:18,240 --> 00:55:22,760 So that is a, that's a, we should all care about that, 965 00:55:22,760 --> 00:55:24,360 right? Humanity needs to care about 966 00:55:24,360 --> 00:55:28,080 where AI is being implemented with regard to national security 967 00:55:29,120 --> 00:55:33,320 challenges, especially at the level where it would have you 968 00:55:33,360 --> 00:55:35,760 know. It could destroy the world, 969 00:55:36,720 --> 00:55:38,160 right? So we. 970 00:55:38,960 --> 00:55:43,800 We conceptually agree with like this responsibility that we have 971 00:55:44,040 --> 00:55:48,160 to ensure that that AI is being implemented ethically, 972 00:55:48,160 --> 00:55:51,080 responsibly, effectively, and that we have communication 973 00:55:51,080 --> 00:55:57,200 between the two countries. I mean, arguably America and 974 00:55:57,200 --> 00:56:00,480 China are kind of the leading voices in AI development. 975 00:56:01,280 --> 00:56:05,200 And so it's important for us to be able to, to have that, you 976 00:56:05,200 --> 00:56:08,200 know, line basically open. And I would love to see that 977 00:56:08,200 --> 00:56:10,320 line of communication open between our senior leaders, 978 00:56:10,400 --> 00:56:11,880 right? Like I would love if the 979 00:56:11,880 --> 00:56:14,760 president had a phone to the president, right? 980 00:56:14,760 --> 00:56:18,520 And they, they could ensure that if something crazy like is going 981 00:56:18,520 --> 00:56:22,640 on, that they, they could talk through it and make sure that it 982 00:56:22,640 --> 00:56:27,560 wasn't some type of, you know, non state actor trying to 983 00:56:27,560 --> 00:56:32,120 influence one party against the other using AI in some fashion 984 00:56:32,120 --> 00:56:34,880 or deception that is powered by AI or right, like there's, 985 00:56:35,080 --> 00:56:37,160 there's real concern here. And so I think it's important 986 00:56:37,160 --> 00:56:41,920 that we have that conversation. And I, I very much think we're, 987 00:56:41,920 --> 00:56:45,360 we're aligned as we're talking to them about this. 988 00:56:46,160 --> 00:56:48,800 So that's one thing I wanted to mention. 989 00:56:51,280 --> 00:56:53,680 I think when it comes to. Where we are. 990 00:56:54,200 --> 00:56:58,520 I think the Air Force and yeah, sure, you could say the whole 991 00:56:58,520 --> 00:57:02,760 Department of War, right. But I think the Air Force was 992 00:57:02,760 --> 00:57:06,480 slow to appreciate and recognise large language models. 993 00:57:06,920 --> 00:57:11,400 And I think that we've come around and that has been an 994 00:57:11,600 --> 00:57:17,600 important pivot over the past couple years months really that 995 00:57:17,600 --> 00:57:21,440 have allowed us to be, you know, more effective in that realm and 996 00:57:21,440 --> 00:57:23,960 and getting these tools out to our our operators. 997 00:57:24,560 --> 00:57:30,360 So that's been good. I think we rightfully are are 998 00:57:30,360 --> 00:57:37,880 slowly making progress in AI being some somewhat enabled in 999 00:57:37,880 --> 00:57:41,880 autonomous capability. So that could be, you know, our 1000 00:57:42,440 --> 00:57:47,680 deliberate focused development of AI agents in uncrewed 1001 00:57:47,680 --> 00:57:50,000 systems. So we're doing that work. 1002 00:57:50,040 --> 00:57:52,800 And I think it's been we've been learning a lot. 1003 00:57:54,120 --> 00:57:58,360 And now you also look at like AI at a higher end, like having the 1004 00:57:58,360 --> 00:58:01,440 ability to press that pickle button, like we're not there 1005 00:58:01,440 --> 00:58:03,200 yet. I don't know when we should get 1006 00:58:03,200 --> 00:58:04,480 there. I don't know if we should get 1007 00:58:04,480 --> 00:58:06,600 there because maybe we find something like, well, we really 1008 00:58:06,600 --> 00:58:09,680 need a human to have full control, even if we're in 1009 00:58:09,680 --> 00:58:12,880 contested environments, even if that means like we're going to 1010 00:58:12,880 --> 00:58:14,360 lose some ground or whatever, like we. 1011 00:58:14,520 --> 00:58:19,400 They need to make that decision as a country, as humanity. 1012 00:58:19,600 --> 00:58:23,440 I'm not sure, but the technology is going to be there. 1013 00:58:23,600 --> 00:58:27,840 We're trying to figure out now like how to how do we adopt AI 1014 00:58:28,040 --> 00:58:34,360 capability into these platforms. And you know, I think that that 1015 00:58:34,360 --> 00:58:38,040 we're making progress there. I don't want to, I don't want to 1016 00:58:38,040 --> 00:58:40,800 get into too many details, but I also don't want people to be 1017 00:58:40,800 --> 00:58:45,880 like scared that the military is like all about AI agents flying 1018 00:58:45,960 --> 00:58:47,920 about these aircraft. Like it is not. 1019 00:58:48,080 --> 00:58:53,720 It is autonomous functions with aspects of that autonomy being 1020 00:58:53,800 --> 00:58:56,240 AI enabled. And it's in the development 1021 00:58:56,240 --> 00:58:58,040 phase. So we are learning about how we 1022 00:58:58,040 --> 00:59:03,800 could possibly do this. I'm, I'm more interested in how 1023 00:59:03,800 --> 00:59:10,800 are we using AI to inform our decision makers about the battle 1024 00:59:10,800 --> 00:59:12,680 space. So that's from automatic target 1025 00:59:12,680 --> 00:59:18,440 recognition to summarization. Like that's just large language 1026 00:59:18,440 --> 00:59:22,840 models to image recreation and image recreation is very 1027 00:59:22,840 --> 00:59:25,240 powerful. So for instance, we don't have 1028 00:59:25,360 --> 00:59:28,120 weather radar that covers the whole world. 1029 00:59:28,440 --> 00:59:31,760 We have weather radar that covers certain aspects of, you 1030 00:59:31,760 --> 00:59:35,200 know, in America, at least in America and on our coast, but 1031 00:59:35,200 --> 00:59:37,680 not everywhere. And so is there a way for us to 1032 00:59:37,680 --> 00:59:42,960 use AI to bring in data from all of these weather satellites from 1033 00:59:42,960 --> 00:59:46,200 around the world to give us what a picture would possibly look 1034 00:59:46,200 --> 00:59:48,320 like about what that weather is? Yeah, right. 1035 00:59:48,320 --> 00:59:51,160 I think we could. So, you know, we're bringing in 1036 00:59:51,160 --> 00:59:55,800 these type of, you know, technologies and, and how do we 1037 00:59:55,800 --> 00:59:58,480 do that to make sure that once again, we're not getting, you 1038 00:59:58,480 --> 00:59:59,360 know, wrong. Information. 1039 00:59:59,360 --> 01:00:04,520 Poison information. Our adversaries, I, I think we 1040 01:00:04,520 --> 01:00:10,600 just need to be aware that if it's technologically possible to 1041 01:00:10,600 --> 01:00:12,800 do the things that I'm like, I'm not sure if we should do this, 1042 01:00:13,000 --> 01:00:14,680 but it's technologically possible to do it. 1043 01:00:14,920 --> 01:00:19,280 We have to assume that someone out there, and I'm not like 1044 01:00:19,280 --> 01:00:22,040 pointing fingers necessarily because it could be a non state 1045 01:00:22,040 --> 01:00:25,600 actor, but someone out there would be willing to do that, 1046 01:00:26,080 --> 01:00:27,600 right. So how do we mitigate that? 1047 01:00:27,680 --> 01:00:31,960 We may not be willing to put like robot dogs onto the battle 1048 01:00:31,960 --> 01:00:34,640 space. They can go shoot a gun, right, 1049 01:00:34,680 --> 01:00:37,560 But our adversary may and adversary I'm not. 1050 01:00:37,720 --> 01:00:38,640 I'm once. Again, I'm not talking. 1051 01:00:38,640 --> 01:00:40,920 About China, I'm just saying like anyone that wants to fight 1052 01:00:40,920 --> 01:00:44,160 us, right, So how do we make sure that we're also prepared 1053 01:00:44,160 --> 01:00:47,520 for that type of fight I think is important. 1054 01:00:48,920 --> 01:00:53,760 Siga, how much time you got? It's 10/2. 1055 01:00:54,240 --> 01:00:57,000 Oh yeah, I have another half hour. 1056 01:00:57,320 --> 01:01:00,000 Easy. OK, Yeah, in that case, let me 1057 01:01:00,000 --> 01:01:01,760 let me just ask you a couple more questions and then I'd like 1058 01:01:01,760 --> 01:01:03,680 to get on to, I'd like to get on talking about the book, but 1059 01:01:04,360 --> 01:01:08,840 yeah. So, so you flew the F35 and 1060 01:01:08,840 --> 01:01:12,840 you've referenced a couple of use cases for AII. 1061 01:01:12,840 --> 01:01:17,840 Wondered if you could tell us two things, what would your top 1062 01:01:18,080 --> 01:01:21,400 application for AI as an F35 guy have been? 1063 01:01:22,720 --> 01:01:25,880 And then also and perhaps these these two are related to each 1064 01:01:25,880 --> 01:01:31,200 other, Where would something like the loyal Wingman programme 1065 01:01:31,200 --> 01:01:35,120 be without AI? Could you do loyal wingman with 1066 01:01:35,120 --> 01:01:38,120 just autonomy? So 2 questions, but maybe 1067 01:01:38,120 --> 01:01:39,840 they're. Related to it, Yeah. 1068 01:01:39,880 --> 01:01:43,280 OK, so remind me too as I get through it. 1069 01:01:43,600 --> 01:01:44,640 You know, if I missed one of them. 1070 01:01:45,760 --> 01:01:51,640 Yeah, For the first one, looking at AI capability now and knowing 1071 01:01:51,640 --> 01:02:00,120 the jet, I think that it would be like I'm more of a take baby 1072 01:02:00,120 --> 01:02:04,440 steps and adopt AI in ways that are more predictable and they 1073 01:02:04,440 --> 01:02:07,360 don't have major consequences type of guy, right. 1074 01:02:07,360 --> 01:02:11,600 So I think using we have voice recognition in the F35, it's not 1075 01:02:11,600 --> 01:02:14,040 very good. It's not like Siri, which also 1076 01:02:14,040 --> 01:02:18,120 gets things wrong, of course, or Alexa, you know, but it is a 1077 01:02:18,120 --> 01:02:20,920 capability. We wouldn't rarely use it and it 1078 01:02:20,920 --> 01:02:23,000 wouldn't understand certain accents, which is always funny. 1079 01:02:23,000 --> 01:02:25,480 Like the test files from Kentucky, like they never 1080 01:02:25,480 --> 01:02:30,280 understood them. But we have that. 1081 01:02:30,280 --> 01:02:33,000 I would love to see that advanced to a degree where I 1082 01:02:33,000 --> 01:02:38,200 could do things with my voice using, you know, speech to text 1083 01:02:39,280 --> 01:02:43,720 kind of interaction and AI could be helpful for that, right? 1084 01:02:43,720 --> 01:02:46,920 Like that's not all AI, but AI could be helpful because it 1085 01:02:46,920 --> 01:02:49,000 could train on someone's voice and it could understand just 1086 01:02:49,000 --> 01:02:53,160 like Alexa trains on certain voices and and those like those 1087 01:02:53,160 --> 01:02:55,640 are the type of applications that I think would be great like 1088 01:02:55,640 --> 01:02:57,400 adopted. It's not going to have major 1089 01:02:57,400 --> 01:02:59,680 implications. It's just going to do this one 1090 01:02:59,680 --> 01:03:03,320 function for me. I think that there's also an 1091 01:03:03,320 --> 01:03:10,320 ability for AI in an F35 to more fully understand what is the 1092 01:03:10,320 --> 01:03:15,560 most effective way to do a tactic by using all of the data 1093 01:03:15,560 --> 01:03:19,680 from our training missions and saying, OK, hey, this is 1094 01:03:19,880 --> 01:03:23,480 actually the way that you should be doing this tactic compared to 1095 01:03:23,480 --> 01:03:24,680 the way that you're doing it now. 1096 01:03:25,400 --> 01:03:28,160 I think it could be very beneficial in that sense because 1097 01:03:28,160 --> 01:03:31,080 that's what it's good at. And maybe it would illuminate A 1098 01:03:31,080 --> 01:03:33,760 tactic or a path that we hadn't thought of before, like, oh, 1099 01:03:33,760 --> 01:03:37,440 I've always been, you know, approaching the enemy when I'm 1100 01:03:37,440 --> 01:03:40,360 at, you know, 20 miles apart from my wingman. 1101 01:03:41,200 --> 01:03:43,920 But now I should be doing it at at 50 feet from me. 1102 01:03:43,920 --> 01:03:45,120 I mean, I'm just making this up, but right. 1103 01:03:45,120 --> 01:03:47,880 Like it could possibly say, hey, do this differently, dummy. 1104 01:03:48,880 --> 01:03:50,120 So I think that would be interesting. 1105 01:03:50,400 --> 01:03:53,280 And I think also the adoption of autonomy that's beyond life 1106 01:03:53,280 --> 01:03:56,720 saving technology, life saving autonomy, life saving autonomy 1107 01:03:56,720 --> 01:03:58,800 is remarkable. But those were kind of baby 1108 01:03:58,800 --> 01:04:00,160 steps. When we talk about automatic 1109 01:04:00,160 --> 01:04:02,080 ground collision avoidance and automatic air collision 1110 01:04:02,080 --> 01:04:05,560 avoidance, those are baby steps to like offensive and defensive 1111 01:04:05,560 --> 01:04:09,240 autonomous capability. Like I want to fly a perfect dog 1112 01:04:09,240 --> 01:04:12,760 fighting scenario right now. Like can I turn on an AI agent 1113 01:04:12,760 --> 01:04:15,040 that helps me optimise my performance? 1114 01:04:15,200 --> 01:04:18,320 Or even like can I turn on an AI agent right now to allow me to 1115 01:04:18,320 --> 01:04:21,840 air refuel, you know, and and get that done very quickly. 1116 01:04:22,560 --> 01:04:27,240 That type of autonomy I think is going to require some AI and 1117 01:04:27,240 --> 01:04:30,320 those would be good implicate those would be good applications 1118 01:04:30,320 --> 01:04:34,360 because the human is still in control, just like auto G cast 1119 01:04:34,360 --> 01:04:36,960 or auto Acas. I can I can turn that off, but I 1120 01:04:36,960 --> 01:04:43,240 can use it to help me, you know, execute in in certain places. 1121 01:04:43,320 --> 01:04:46,600 So I think that's what I would think about when I'm looking to 1122 01:04:46,680 --> 01:04:49,720 like try to adopt AI solutions into the F35. 1123 01:04:50,840 --> 01:04:52,920 What was your second question? Loyal wingman. 1124 01:04:53,040 --> 01:04:55,400 So thank you. Could loyal wingman work with 1125 01:04:55,400 --> 01:04:57,720 just a a bunch of hard coded autonomy? 1126 01:04:57,760 --> 01:05:01,960 Absolutely, yeah. Loyal wingman could work to a 1127 01:05:01,960 --> 01:05:07,000 degree without any type of AII mean right now. 1128 01:05:07,000 --> 01:05:10,320 Like you'd argue that that is like what we're developing. 1129 01:05:10,520 --> 01:05:14,080 They're not all like the AI agent that we're testing in the 1130 01:05:14,080 --> 01:05:19,320 XQ 58 like we're doing other autonomy without AI agents. 1131 01:05:19,320 --> 01:05:22,320 So, yeah, absolutely. And could it still be effective? 1132 01:05:22,320 --> 01:05:25,880 Yeah, absolutely. I think the problem becomes in 1133 01:05:25,880 --> 01:05:31,760 certain scenarios that those autonomous systems will not be 1134 01:05:31,760 --> 01:05:36,040 as effective because they won't be able to deal with all the ifs 1135 01:05:36,560 --> 01:05:38,440 compared to all the then's, right? 1136 01:05:38,440 --> 01:05:41,600 Like the if then statements because it's just really complex 1137 01:05:41,600 --> 01:05:45,520 if you you're not going to be able to hard code into the 1138 01:05:45,520 --> 01:05:49,440 software all of the possible things that it could run into, 1139 01:05:50,080 --> 01:05:52,160 run into. I just mean like experience, 1140 01:05:52,160 --> 01:05:55,920 like maybe different signals or different, you know, aspects of 1141 01:05:55,920 --> 01:05:57,560 the sensor where maybe. It. 1142 01:05:58,440 --> 01:06:01,160 Doesn't have hard coded that if it gets this one signal that's 1143 01:06:01,160 --> 01:06:04,040 an enemy threat and it doesn't know that, so it doesn't respond 1144 01:06:04,040 --> 01:06:07,840 correctly when an AI agent could learn like, oh, that's a signal 1145 01:06:07,840 --> 01:06:11,600 that I don't know, but it's similar to this other signal. 1146 01:06:11,720 --> 01:06:14,880 And so I'm going to respond to it like it's in a threat, you 1147 01:06:14,880 --> 01:06:18,560 know, but the deterministic code may not, you know, be able to be 1148 01:06:18,560 --> 01:06:21,320 as responsive. So I think that that is what you 1149 01:06:21,320 --> 01:06:23,400 could absolutely do. You're just going to have 1150 01:06:23,400 --> 01:06:26,000 certain limitations and you're going to have to hard code into 1151 01:06:26,000 --> 01:06:29,200 it like if you lose communication with if the pilot 1152 01:06:29,360 --> 01:06:32,080 do these things, but that becomes complicated again, 1153 01:06:32,720 --> 01:06:34,560 right? So it's going to be limiting, 1154 01:06:35,360 --> 01:06:42,080 but not, not detrimental or you know, complete barrier to the 1155 01:06:42,080 --> 01:06:45,680 capability A. Final question for me on this 1156 01:06:45,680 --> 01:06:47,760 then and then I'd like to ask a couple of questions from my 1157 01:06:47,760 --> 01:06:49,280 audience. I have subscribers and one of 1158 01:06:49,280 --> 01:06:51,520 the same same as the final pilot podcast. 1159 01:06:52,040 --> 01:06:54,320 They can ask questions and I'll authorise it and then we'll, 1160 01:06:54,640 --> 01:06:56,040 we'll talk about your, your book. 1161 01:06:56,040 --> 01:06:59,680 But one of the things that you're seeing in, in discourse 1162 01:06:59,680 --> 01:07:04,360 today around commercial use of AI is deskilling of people. 1163 01:07:05,080 --> 01:07:08,840 You know, if you, let's say you use AI to write your website for 1164 01:07:08,840 --> 01:07:10,920 you and your marketing brochures or whatever, then you're 1165 01:07:10,920 --> 01:07:13,400 deskilling effectively. Or you're, the risk is you end 1166 01:07:13,400 --> 01:07:16,680 up deskilling your marketing workforce because then they 1167 01:07:16,680 --> 01:07:18,440 never learn how to write the brochures or whatever. 1168 01:07:18,440 --> 01:07:23,720 That's a just a random example. Does the same apply to the fast 1169 01:07:23,720 --> 01:07:25,640 jet community, to the tactical operator? 1170 01:07:26,120 --> 01:07:32,280 The idea that if AI becomes so embedded and entrenched in the 1171 01:07:32,280 --> 01:07:37,640 process of employing that the he or she will as the operator will 1172 01:07:37,640 --> 01:07:41,200 actually not have the ability to make judgments, not have the 1173 01:07:41,200 --> 01:07:46,440 skills and the experience and the ability to draw on their own 1174 01:07:46,440 --> 01:07:49,120 knowledge in order to determine what the correct course of 1175 01:07:49,120 --> 01:07:51,160 action is. And then if they've got the 1176 01:07:51,160 --> 01:07:53,280 hammer on the, you know, they've got the button on the, the 1177 01:07:53,280 --> 01:07:56,360 pickle button control of the pickle button, then their 1178 01:07:56,360 --> 01:07:58,800 judgement for for hitting that pickle button at the end of the 1179 01:07:58,800 --> 01:08:01,160 day may not be what it should be. 1180 01:08:01,160 --> 01:08:03,680 Is that a realistic or probabilistic thing? 1181 01:08:05,760 --> 01:08:10,840 I think your listeners and our society writ large need to 1182 01:08:10,840 --> 01:08:14,520 understand that there's always going to be disruptors to our 1183 01:08:14,520 --> 01:08:18,399 professions and creative destruction is going to happen 1184 01:08:18,399 --> 01:08:22,760 and is not a bad thing. It can be in the short term 1185 01:08:23,120 --> 01:08:28,399 detrimental to someone's life. Like it may impact their their 1186 01:08:28,399 --> 01:08:32,880 immediate work, it may impact prospects that they had. 1187 01:08:33,880 --> 01:08:37,720 So I think in general, it's good for us to accept like the 1188 01:08:37,720 --> 01:08:41,600 responsibility that we have to continue continually push 1189 01:08:41,600 --> 01:08:45,920 ourselves forward to learn about what's coming, to figure out how 1190 01:08:45,920 --> 01:08:49,080 that is going to play a part in our world, our profession, 1191 01:08:49,080 --> 01:08:54,520 whatever it may be, and to invest our time and our energy 1192 01:08:54,680 --> 01:09:01,359 into figuring out like, So what, like how is this going to be a 1193 01:09:01,359 --> 01:09:03,560 player for me? And then how am I going to 1194 01:09:03,560 --> 01:09:06,680 address it? So the answer to your question 1195 01:09:06,680 --> 01:09:10,720 is absolutely like it is going to have an impact, but how 1196 01:09:10,720 --> 01:09:13,040 detrimental or bad is that going to be? 1197 01:09:13,279 --> 01:09:15,359 I don't know, but I don't think any of us know. 1198 01:09:15,520 --> 01:09:21,279 But I do know through other things that have happened in our 1199 01:09:21,279 --> 01:09:24,439 society. So from the ATM example is 1200 01:09:24,439 --> 01:09:27,359 always a classic one. Like ATMs did not stop us 1201 01:09:27,359 --> 01:09:30,439 needing bankers, though they thought before they adopted them 1202 01:09:30,439 --> 01:09:31,600 that that was what was going to happen. 1203 01:09:32,240 --> 01:09:34,960 And you can look at like lots of different, you know, careers 1204 01:09:34,960 --> 01:09:37,960 that people thought was going to be impacted, though wasn't 1205 01:09:37,960 --> 01:09:42,000 necessarily it was impacted, but it didn't mean it was gone 1206 01:09:42,000 --> 01:09:44,640 forever. But I also think of aviation and 1207 01:09:44,640 --> 01:09:50,640 how it has kind of through the years changed what it means to 1208 01:09:50,640 --> 01:09:52,840 be a changed what it means to be a fighter pilot. 1209 01:09:53,720 --> 01:09:55,440 And I was having this conversation with someone 1210 01:09:55,440 --> 01:10:00,680 recently too, like, I think back before I had, you know, a God's 1211 01:10:00,680 --> 01:10:02,720 eye view. So, you know, as if God were 1212 01:10:02,720 --> 01:10:06,160 looking down and I had an image of where I am on a map, right? 1213 01:10:06,160 --> 01:10:10,240 Like we would have to figure out where we were in our brains by 1214 01:10:10,240 --> 01:10:13,120 cutting radials. I'd be like, OK, I'm at a 180 1215 01:10:13,120 --> 01:10:15,520 radial at 20 miles off of this frequency. 1216 01:10:15,680 --> 01:10:17,560 I need to like make sure I tune that frequency. 1217 01:10:17,560 --> 01:10:19,600 That's the right frequency. This is the right radial. 1218 01:10:19,800 --> 01:10:21,760 And then we'd have to like create a picture in our own 1219 01:10:21,760 --> 01:10:25,080 mind. And you bring in a, a picture, a 1220 01:10:25,080 --> 01:10:28,040 moving map where like you have the plane on the on the picture 1221 01:10:28,040 --> 01:10:29,520 and you're like, OK, that's where I am. 1222 01:10:29,920 --> 01:10:35,880 Did that, did that change a pilot's ability to perceive in 1223 01:10:35,880 --> 01:10:39,840 their own brain where they were in relation to Avor or tack an 1224 01:10:40,040 --> 01:10:41,440 using radios? Yeah, yeah. 1225 01:10:41,440 --> 01:10:45,920 Did that matter? Like no, like, like is the same 1226 01:10:45,920 --> 01:10:50,160 with data links. So now I have a, a picture of me 1227 01:10:50,200 --> 01:10:52,640 on a moving map and I bring in data links so I'm able to 1228 01:10:52,640 --> 01:10:56,320 communicate with my wingman and I'm able to see my wingman and 1229 01:10:56,320 --> 01:11:02,040 I'm able to effectively manage kind of our formation from this 1230 01:11:02,040 --> 01:11:05,880 moving map. I've, I've noticed that because 1231 01:11:05,880 --> 01:11:08,400 I was, I was here as we are transitioning to all of this 1232 01:11:08,400 --> 01:11:10,760 technology. I was in the fighter cockpit 1233 01:11:10,760 --> 01:11:15,120 doing it and I noticed that, yeah, pilots weren't as good at 1234 01:11:15,120 --> 01:11:17,880 certain things. They weren't as aware of 1235 01:11:18,280 --> 01:11:20,680 listening to the radio and being able to build the picture out 1236 01:11:20,680 --> 01:11:23,640 more effectively and they relied too much on the moving map and 1237 01:11:23,640 --> 01:11:26,000 the data links. But was that a bad thing? 1238 01:11:26,000 --> 01:11:27,800 Like, no. Fast forward a few years 1239 01:11:27,880 --> 01:11:30,240 transition like they're not as good at that, but they're good 1240 01:11:30,240 --> 01:11:34,440 at other things, right? And so now you bring in you 1241 01:11:34,440 --> 01:11:38,200 know, additional technology and you say, OK, I have an auto 1242 01:11:38,200 --> 01:11:41,240 throttle and I have an autopilot and I have an aircraft that's 1243 01:11:41,240 --> 01:11:45,280 really easy to land in the F35. Like what does it mean to be a 1244 01:11:45,440 --> 01:11:47,760 fighter pilot now? It's different than if I'm 1245 01:11:47,760 --> 01:11:52,840 flying AF15 than if I'm flying an F4, that if I'm flying AP 51, 1246 01:11:52,840 --> 01:11:55,600 right? And so we have to be able to, 1247 01:11:55,840 --> 01:11:58,840 you know, kind of grow with these these opportunities, these 1248 01:11:58,840 --> 01:12:01,200 changes. And yeah, there's certain things 1249 01:12:01,200 --> 01:12:06,640 that are worse, like pilots are going to be worse at landing if 1250 01:12:07,600 --> 01:12:09,480 if. If the system's always landing 1251 01:12:09,480 --> 01:12:11,160 for them. But at what? 1252 01:12:11,160 --> 01:12:14,800 Point is that going to be like dangerous or is it just like 1253 01:12:14,800 --> 01:12:19,880 this transition period to be able to adopt technology that is 1254 01:12:19,880 --> 01:12:23,320 going to be more effective? And yeah, it's classic like, OK, 1255 01:12:23,320 --> 01:12:27,680 well you take airline pilots and you replace them with computers 1256 01:12:27,840 --> 01:12:32,480 if or you augment them with computers, are are those airline 1257 01:12:32,480 --> 01:12:37,160 pilots going to be now less safe because they don't actually know 1258 01:12:37,160 --> 01:12:40,400 some of the things maybe to an extent, but maybe not maybe a 1259 01:12:40,400 --> 01:12:43,040 holistically they're actually going to be safer because the 1260 01:12:43,040 --> 01:12:46,080 combination of the two, the airline pilot with the 1261 01:12:46,120 --> 01:12:49,320 autonomous system or software AI agent, whatever actually make 1262 01:12:49,320 --> 01:12:53,760 them more effective. And so I, I think it's right for 1263 01:12:53,760 --> 01:12:57,320 us to be concerned and we need to think through that. 1264 01:12:57,320 --> 01:13:02,000 But I also, I think I'm more optimistic that we are going to 1265 01:13:02,000 --> 01:13:04,920 be able to work alongside this and we're going to be able to 1266 01:13:04,960 --> 01:13:07,240 create things that we didn't know and things are going to 1267 01:13:07,440 --> 01:13:11,160 change and in really cool ways. While at the same time we have 1268 01:13:11,160 --> 01:13:14,360 to address that there are going to be certain jobs that are 1269 01:13:14,360 --> 01:13:15,680 going to be changed because of it. 1270 01:13:15,920 --> 01:13:17,960 And it's like, you know, software coding is probably 1271 01:13:17,960 --> 01:13:20,480 never going to be the same. Marketing is probably never 1272 01:13:20,480 --> 01:13:22,280 going to be the same. Does it mean we're not going to 1273 01:13:22,280 --> 01:13:24,880 be marketing? No, I think we're just going to 1274 01:13:25,120 --> 01:13:28,040 be adopting this. Book editing and book publishing 1275 01:13:28,040 --> 01:13:29,400 is probably never going to be the same. 1276 01:13:29,800 --> 01:13:34,160 Right, like I will tell you, I told you I used editorial AI 1277 01:13:34,160 --> 01:13:41,880 tools for my book and when I gave my face like how few? 1278 01:13:41,880 --> 01:13:43,920 Errors there were. And it's not because I knew what 1279 01:13:43,920 --> 01:13:49,520 I was doing. It's because use is instead of 1280 01:13:49,520 --> 01:13:51,400 our. You know, wrong person here, 1281 01:13:51,400 --> 01:13:53,680 right? And so we can see that it is 1282 01:13:53,680 --> 01:13:56,520 changing it, but I think the editors is probably going to 1283 01:13:56,520 --> 01:13:58,360 have to change maybe their relationship. 1284 01:13:58,360 --> 01:14:00,040 With the the. Writer is going to be different 1285 01:14:00,040 --> 01:14:02,280 and maybe more advanced. Maybe they could help in certain 1286 01:14:02,280 --> 01:14:04,920 ways that now they have time to do that. 1287 01:14:05,520 --> 01:14:11,040 And so clearly very important topic, but I'm probably more on 1288 01:14:11,040 --> 01:14:14,040 the optimistic side that I think we can do this together because 1289 01:14:14,040 --> 01:14:17,480 I think humanity appreciates like working with a human. 1290 01:14:17,480 --> 01:14:19,000 And so we're not trying to stop that. 1291 01:14:19,000 --> 01:14:26,200 We're just trying to enhance it. OK, Cinco, let me ask you a 1292 01:14:26,200 --> 01:14:29,440 couple of questions from my audience then just give me a 1293 01:14:29,440 --> 01:14:30,880 SEC. Here we go then. 1294 01:14:30,880 --> 01:14:33,080 So this, this is just a quick couple of questions then for 1295 01:14:33,080 --> 01:14:37,200 you, Cinco, from my audience and there were, there were quite a 1296 01:14:37,200 --> 01:14:38,560 few. So I'm just going to cut them 1297 01:14:38,560 --> 01:14:41,840 down to just one or two for you. So, but Knox asked, with the 1298 01:14:41,840 --> 01:14:45,400 rapid advancement of open source models across various ML 1299 01:14:45,400 --> 01:14:49,760 disciplines, how much has the US Air Force been able to benefit? 1300 01:14:50,640 --> 01:14:53,680 And is there a requirement to author original models? 1301 01:14:55,720 --> 01:15:01,280 I think I think it's a benefit in general for these models to 1302 01:15:01,280 --> 01:15:06,080 be accessible though. I am like empathetic for sure 1303 01:15:07,000 --> 01:15:10,440 and mainly for security reasons of how these models we need to 1304 01:15:10,440 --> 01:15:13,200 be mindful of the weights that these companies are using 1305 01:15:14,120 --> 01:15:17,440 because that is not just proprietary, but I think can be 1306 01:15:17,440 --> 01:15:21,160 then manipulated to a way that would change the models in bad 1307 01:15:21,160 --> 01:15:25,120 ways. And so I I think one, it is 1308 01:15:25,200 --> 01:15:28,200 valuable to be able to communicate effectively and have 1309 01:15:28,200 --> 01:15:31,640 open source to a degree. As long as we are considering 1310 01:15:31,640 --> 01:15:34,640 non state actors and people that are really trying to do horrible 1311 01:15:34,640 --> 01:15:40,360 things throughout the world. The Air Force, I think in the 1312 01:15:40,360 --> 01:15:44,120 military would benefit from having instantiations that are 1313 01:15:44,120 --> 01:15:49,680 separated from that type of pipeline that goes back to a 1314 01:15:49,680 --> 01:15:52,200 company for many reasons. One, for security reasons, 1315 01:15:53,800 --> 01:15:57,520 mainly for security reasons. And then I think we need to 1316 01:15:57,520 --> 01:16:04,720 manage it so that we're able to ensure that we have more 1317 01:16:04,720 --> 01:16:08,720 governance over the systems. And I'm, I'm trying to think of 1318 01:16:08,720 --> 01:16:11,160 the word, but we basically want to make sure that we have 1319 01:16:11,160 --> 01:16:17,600 control over what is training a system and that how our inputs 1320 01:16:17,600 --> 01:16:21,160 or our questions are being protected so that, you know, 1321 01:16:21,200 --> 01:16:25,360 like you can learn a lot from if you go into someone's ChatGPT 1322 01:16:25,760 --> 01:16:28,920 query and figure out like what they're asking you, just like 1323 01:16:28,920 --> 01:16:32,480 you would learn a lot from going into Google's search bar. 1324 01:16:32,720 --> 01:16:37,360 And if the, you know, if the US government's employees are 1325 01:16:37,360 --> 01:16:40,720 searching, like what happens when I ran as, you know, like 1326 01:16:41,000 --> 01:16:43,480 you're giving, you're giving away information. 1327 01:16:43,680 --> 01:16:47,280 My point is, I'm not sure when this will air, but of course 1328 01:16:47,280 --> 01:16:50,080 we're we're closely watching what's going on in the Middle 1329 01:16:50,080 --> 01:16:51,720 East right now. But I just was using that as an 1330 01:16:51,720 --> 01:16:55,840 example of like, we don't want to give away information that is 1331 01:16:55,840 --> 01:16:59,760 not, you know, not to be given away. 1332 01:16:59,760 --> 01:17:02,920 Put it that way. So Matthew has asked. 1333 01:17:03,320 --> 01:17:06,120 So watching current conflicts, it sometimes appears as though 1334 01:17:06,120 --> 01:17:10,400 the technology becomes more and more advanced, but the most 1335 01:17:10,400 --> 01:17:12,920 effective countermeasures are becoming more and more 1336 01:17:12,920 --> 01:17:16,200 rudimentary. At what point does AI start 1337 01:17:16,400 --> 01:17:21,160 producing diminishing returns? And in the AI digital world, how 1338 01:17:21,160 --> 01:17:23,200 do you defend against analogue threats? 1339 01:17:23,200 --> 01:17:25,360 That's a really interesting question, I think. 1340 01:17:29,000 --> 01:17:32,760 I think what we're seeing is a big transition going on in 1341 01:17:32,760 --> 01:17:40,200 different careers in different, different aspects of of society. 1342 01:17:41,400 --> 01:17:45,280 And why I say that is, I do think that, yes, you may find 1343 01:17:45,480 --> 01:17:50,680 that there's ways to counter using like legacy or analogue 1344 01:17:50,680 --> 01:17:54,440 approaches, but I don't think that those will actually last 1345 01:17:54,440 --> 01:18:00,480 because I think AI can be taught or learn to effectively address 1346 01:18:00,480 --> 01:18:04,240 that as it matures. And so I don't want to be 1347 01:18:04,320 --> 01:18:08,200 doomsday or anything. I'm just saying like, yeah, the 1348 01:18:08,200 --> 01:18:14,040 best example or an example of maybe not the best, this idea of 1349 01:18:14,040 --> 01:18:17,640 a Centaur. So it, it's kind of like, you 1350 01:18:17,640 --> 01:18:24,520 know, half human, half horse. AI beat Garry Kasparov at chess 1351 01:18:24,760 --> 01:18:30,520 in what, 97 or something. And over the ensuing years that 1352 01:18:30,520 --> 01:18:35,320 followed, AI was very good at chess and humans were good at 1353 01:18:35,320 --> 01:18:37,280 chess, and AI could beat a human. 1354 01:18:37,680 --> 01:18:42,320 But if an AI and a human worked together, and it was for a long 1355 01:18:42,320 --> 01:18:45,960 period of time, you know, over a decade, if they work together, 1356 01:18:46,280 --> 01:18:50,720 they would together both beat an AI agent or together both beat a 1357 01:18:50,720 --> 01:18:56,920 human. But then as AI continued to go 1358 01:18:58,000 --> 01:19:01,680 at the speed in which it's going, it would no longer need a 1359 01:19:01,680 --> 01:19:04,040 human at all. It just is able to beat 1360 01:19:04,040 --> 01:19:06,680 everyone. Like there's no chance a human 1361 01:19:06,680 --> 01:19:09,400 has. So it's a Centaur. 1362 01:19:09,760 --> 01:19:12,400 It's a transition. Like, at what point are you the 1363 01:19:12,400 --> 01:19:14,560 Centaur together or that you don't need? 1364 01:19:14,720 --> 01:19:17,800 And I do feel like if we have a mentality of, oh, we're going to 1365 01:19:17,800 --> 01:19:21,640 be able to defeat AI by doing this, that that is not 1366 01:19:21,640 --> 01:19:24,000 necessarily going to be true in a year or two years, five years, 1367 01:19:24,000 --> 01:19:28,600 10 years from now. We are just starting the AI 1368 01:19:28,600 --> 01:19:33,600 revolution. It is really, I mean, I really 1369 01:19:33,600 --> 01:19:36,240 believe that it's the tip of an iceberg, like we're just 1370 01:19:36,240 --> 01:19:38,880 scraping out because large language models are cute and 1371 01:19:38,880 --> 01:19:40,840 that's what we've been living in for the past few years. 1372 01:19:41,200 --> 01:19:46,040 But when AI really meets physical autonomous systems, 1373 01:19:47,120 --> 01:19:49,600 that's going to change lots of things. 1374 01:19:49,920 --> 01:19:52,480 So I don't think we're there yet fully. 1375 01:19:52,640 --> 01:19:56,000 We're starting to see it with cars, right? 1376 01:19:56,000 --> 01:19:57,760 That's why Tesla's able to be effective. 1377 01:19:58,000 --> 01:20:00,880 And that's going to change quite a bit, especially as cyber taxi, 1378 01:20:00,880 --> 01:20:03,560 cyber cab come out soon. I think that's going to change 1379 01:20:03,560 --> 01:20:06,920 when Optimist comes out. And I'm not like here as a Tesla 1380 01:20:06,920 --> 01:20:09,120 commercial, but you know, those are kind of the best examples 1381 01:20:09,120 --> 01:20:13,160 that people see. I think AI with the physical 1382 01:20:13,160 --> 01:20:16,440 systems is going to actually revolutionise even more and not 1383 01:20:16,440 --> 01:20:19,640 just like doing your dishes for you or folding your laundry, but 1384 01:20:19,640 --> 01:20:24,560 like having a meaningful, you know, part of our medical 1385 01:20:24,560 --> 01:20:28,920 fields. Our some of our blue collar work 1386 01:20:28,960 --> 01:20:31,600 that is going on from like stocking warehouses to moving 1387 01:20:31,600 --> 01:20:35,680 things to driving vehicles or right like there's going to be a 1388 01:20:35,680 --> 01:20:37,920 lot of that. And so I don't think we're at 1389 01:20:37,920 --> 01:20:40,760 the point yet where we're worried about a diminishing 1390 01:20:40,760 --> 01:20:42,960 return because we haven't even seen really. 1391 01:20:43,440 --> 01:20:45,840 Like the the. Future that's about to unfold, 1392 01:20:45,840 --> 01:20:50,440 which is pretty remarkable. Scary moments, exciting and 1393 01:20:50,440 --> 01:20:55,960 moments for us to help define. So this question is from Biggles 1394 01:20:55,960 --> 01:20:58,880 Tinton, he says, I know most Western nations have strict 1395 01:20:58,880 --> 01:21:03,120 policies on the usage of AI. You kind of touched on this, but 1396 01:21:04,240 --> 01:21:06,400 you know, I'll ask anyway. Do you think that as in in the 1397 01:21:06,400 --> 01:21:09,440 result of a a major war, a lot of those limitations would be 1398 01:21:09,440 --> 01:21:13,520 reduced to help win? I think that there are certain 1399 01:21:13,520 --> 01:21:16,560 lines we should never cross right? 1400 01:21:16,840 --> 01:21:20,200 And especially with like AI interfacing with nuclear 1401 01:21:20,200 --> 01:21:21,960 decision making, like it just should never happen. 1402 01:21:22,320 --> 01:21:25,320 I mean, nuclear war is a losing proposition for all parties, 1403 01:21:26,000 --> 01:21:28,400 horrible for humanity. And I think there's aspect of an 1404 01:21:28,440 --> 01:21:31,800 AI engagement that could also turn into the same thing, loss 1405 01:21:31,800 --> 01:21:33,840 of control of systems. And those are, these are 1406 01:21:33,840 --> 01:21:37,920 concerns. So I think there's a reality 1407 01:21:37,920 --> 01:21:39,760 that there's certain lines never to be crossed. 1408 01:21:40,680 --> 01:21:43,400 Now that there's this, there's this grey area of like, well, 1409 01:21:43,400 --> 01:21:48,240 would we in a situation do this? And you can you almost have to 1410 01:21:48,680 --> 01:21:52,520 imagine that world of like, well, if we're backed into a 1411 01:21:52,520 --> 01:21:56,080 corner, what would we do? What would we expect from an 1412 01:21:56,080 --> 01:21:58,120 adversary? What would we expect from 1413 01:21:58,120 --> 01:22:01,720 ourselves? And I would hope that we can 1414 01:22:01,720 --> 01:22:09,280 agree to not do certain things. And I think we've as a like a, a 1415 01:22:09,280 --> 01:22:12,920 world decided like that we're not going to use certain types 1416 01:22:12,920 --> 01:22:14,800 of land mines, right? That we're not going to use 1417 01:22:14,800 --> 01:22:17,160 certain type of biological weapons that we're not going to 1418 01:22:17,440 --> 01:22:19,200 right. And so I think we can put 1419 01:22:19,200 --> 01:22:23,040 certain limits on it. I think we need to be very 1420 01:22:23,040 --> 01:22:27,560 mindful of when we're willing to adopt some of the capability. 1421 01:22:28,120 --> 01:22:30,520 We don't want to adopt it when it's immature. 1422 01:22:30,520 --> 01:22:33,520 We don't want to have to be forced into a situation because 1423 01:22:33,520 --> 01:22:36,680 I think that something bad could or would happen. 1424 01:22:37,560 --> 01:22:41,680 And so I think it's a it's important question that I don't 1425 01:22:41,680 --> 01:22:44,760 think there's a clear answer on other than we have to understand 1426 01:22:44,760 --> 01:22:49,640 our own humanity and look at our past and, and appreciate that we 1427 01:22:49,640 --> 01:22:53,800 may be tempted to do something hasty. 1428 01:22:53,960 --> 01:22:56,720 We may be tempted to use a last resort. 1429 01:22:56,720 --> 01:23:01,000 I mean, imagine the entire world is going to be taken over by a 1430 01:23:01,200 --> 01:23:05,160 horrible dictator like Hitler or something, right? 1431 01:23:05,160 --> 01:23:08,560 Like what tools would we be willing to use to stop that from 1432 01:23:08,560 --> 01:23:12,320 happening? And I think we have to try to 1433 01:23:12,520 --> 01:23:16,240 almost think through that now because it would like it's 1434 01:23:16,240 --> 01:23:17,600 important question. I don't know. 1435 01:23:17,600 --> 01:23:20,200 I don't know what the answer is, but I think we have to be real 1436 01:23:20,200 --> 01:23:25,120 with trying to think through it. So, guy, I could carry on 1437 01:23:25,120 --> 01:23:28,280 talking with you about this for, for many more hours and I've got 1438 01:23:28,280 --> 01:23:30,240 loads more questions, but I won't, I won't put you through 1439 01:23:30,240 --> 01:23:32,600 that because that would, that would not give us time to talk 1440 01:23:32,600 --> 01:23:35,000 about your book, which I equally want to discuss with you. 1441 01:23:35,680 --> 01:23:37,800 I really enjoyed reading it. So it's called unlocking the 1442 01:23:37,800 --> 01:23:41,120 last 20% and it draws on, we said right at the beginning of 1443 01:23:41,120 --> 01:23:43,520 this conversation, you talked about it a tiny bit draws on 1444 01:23:43,760 --> 01:23:45,640 many of your experiences going through childhood. 1445 01:23:45,640 --> 01:23:47,840 I, I was really interested to hear you talk about the letter 1446 01:23:47,840 --> 01:23:51,200 you'd, your dad had written. You, you know, you, I think you, 1447 01:23:51,200 --> 01:23:53,880 you titled the chapter Disappointing my father. 1448 01:23:53,880 --> 01:23:56,240 I think that's how you title it. And you go through all the way 1449 01:23:56,240 --> 01:23:59,560 through then being homeschooled, then going to university, 1450 01:23:59,880 --> 01:24:03,400 getting into the Air Force, flying the Eagle, standing up 1451 01:24:03,400 --> 01:24:05,800 the MC 12, and then, and then eventually becoming a test 1452 01:24:05,800 --> 01:24:08,000 pilot. And you illustrate then all of 1453 01:24:08,000 --> 01:24:11,360 these points that you have with experiences that you that taught 1454 01:24:11,360 --> 01:24:14,240 you those things. So I, so I, I guess that the 1455 01:24:14,240 --> 01:24:16,400 first question I wanted to ask you about it was, was what 1456 01:24:16,400 --> 01:24:17,920 inspired you to do it? Why? 1457 01:24:17,960 --> 01:24:23,200 Why did you write it? Well, thank you for one reading 1458 01:24:23,200 --> 01:24:25,360 it. You're one of the few people who 1459 01:24:25,360 --> 01:24:29,560 have read it. I originally started writing 1460 01:24:29,560 --> 01:24:33,080 down stories over a decade ago because I lost my father when I 1461 01:24:33,080 --> 01:24:37,000 was 21 to cancer, and we had a great relationship. 1462 01:24:37,560 --> 01:24:39,800 I learned a lot from him, but there were so many questions I 1463 01:24:39,800 --> 01:24:42,200 didn't get to ask him that I didn't even know to ask him 1464 01:24:42,360 --> 01:24:46,200 because I was so young. And so I started writing things 1465 01:24:46,200 --> 01:24:51,240 down in order to capture some of my thoughts and perspectives for 1466 01:24:51,240 --> 01:24:54,040 my children. And then as it kind of coalesced 1467 01:24:54,040 --> 01:24:58,040 and came together, I felt like, you know, some of my stories are 1468 01:24:58,320 --> 01:25:02,560 pretty crazy, and a lot of them are failures, by the way. 1469 01:25:03,520 --> 01:25:08,160 But I'm still here and I, I figured there was maybe 1470 01:25:08,160 --> 01:25:11,080 something in the way that I think about life or that I've 1471 01:25:11,080 --> 01:25:13,680 gone through life or experienced certain things that that could 1472 01:25:13,920 --> 01:25:16,440 build on someone's perspectives to help them deal with 1473 01:25:16,440 --> 01:25:19,200 challenges that they're facing, a failure that they're facing, 1474 01:25:19,200 --> 01:25:20,680 or even like what is their purpose? 1475 01:25:20,680 --> 01:25:25,800 And how do they kind of approach becoming the person they want to 1476 01:25:25,800 --> 01:25:26,840 become? Right. 1477 01:25:26,840 --> 01:25:31,960 And, and so that was kind of the catalyst was my kids and turned 1478 01:25:31,960 --> 01:25:35,680 into an opportunity, I think, to, to share stories and, and 1479 01:25:35,680 --> 01:25:41,240 try to kind of give a perspective that I think is 1480 01:25:41,240 --> 01:25:43,800 needed in this time and age. Because there's so many 1481 01:25:43,800 --> 01:25:48,160 distractions and there's so many things going on that I think I 1482 01:25:48,160 --> 01:25:51,040 think people are kind of saturated with the story of 1483 01:25:51,040 --> 01:25:54,680 like, you need to pick yourself back up or you need to go like, 1484 01:25:54,680 --> 01:25:57,360 you just need to try harder. Or it's like this individual 1485 01:25:57,360 --> 01:25:59,800 focus. And I start with this individual 1486 01:25:59,800 --> 01:26:02,880 focus of the book, but it's, it's really about like community 1487 01:26:03,160 --> 01:26:05,160 and it's about who we surround ourselves with. 1488 01:26:05,160 --> 01:26:09,280 And it's about love and it's about like resiliency. 1489 01:26:09,520 --> 01:26:14,840 And I, I, I wanted to be able to kind of share that perspective 1490 01:26:15,160 --> 01:26:18,840 to be somewhat of a counterbalance to, I think what 1491 01:26:18,840 --> 01:26:21,480 we often hear, which is just like try harder. 1492 01:26:23,040 --> 01:26:25,240 Yeah, I like it in the, in the sense that it has a lot of 1493 01:26:25,240 --> 01:26:27,640 practicality around it. I mean, you, you talk about 1494 01:26:27,640 --> 01:26:30,800 mindfulness, you describe visualising things. 1495 01:26:30,800 --> 01:26:34,040 I've always found, found that, you know, not, not that I have 1496 01:26:34,040 --> 01:26:37,440 ever unlocked my last 20%, but I've always found that 1497 01:26:37,440 --> 01:26:40,560 visualising myself being something or doing something is, 1498 01:26:40,560 --> 01:26:43,440 is very helpful in, in getting me to, to the point where maybe 1499 01:26:43,440 --> 01:26:47,520 I'll, I'll make that happen. One of the things that does come 1500 01:26:47,520 --> 01:26:50,880 through both in your introduction and then in, at the 1501 01:26:50,880 --> 01:26:54,520 end where you, where you provide acknowledgements is, is the 1502 01:26:54,520 --> 01:26:58,240 importance of, of your faith and your family to you. 1503 01:26:59,640 --> 01:27:04,000 I, I, I read when I was reading the chapter where you talk about 1504 01:27:04,000 --> 01:27:08,960 flying the F15 on an A120 test and you've got to do mark 2.5 1505 01:27:09,000 --> 01:27:12,840 and you've got to push negative 3GS could hold it there for a 1506 01:27:12,840 --> 01:27:15,560 long time. You describe how as a test 1507 01:27:15,560 --> 01:27:18,240 pilot, your job is to go into the unknown. 1508 01:27:20,320 --> 01:27:22,200 And, and so there were two things that struck me about 1509 01:27:22,200 --> 01:27:24,440 that. The first was I thought I would 1510 01:27:24,440 --> 01:27:27,480 have been absolutely petrified. I mean, it, it sounds proper 1511 01:27:27,480 --> 01:27:29,800 scary what you, what you had to do on that test. 1512 01:27:29,800 --> 01:27:34,000 And I'm sure that was just one of many, but then this sort of 1513 01:27:34,800 --> 01:27:37,520 disconnect between doing that and then having this family at 1514 01:27:37,520 --> 01:27:42,000 home and the, the risk for you having it all on the line in 1515 01:27:42,000 --> 01:27:44,920 the, the risk for them as well. And I suppose one of the 1516 01:27:44,920 --> 01:27:47,680 questions I wanted to ask you was, was how you manage that? 1517 01:27:47,680 --> 01:27:52,960 How do you, how do you go through life with a very 1518 01:27:53,600 --> 01:27:58,000 obviously a very ordered and logical mind and the ability to 1519 01:27:58,000 --> 01:28:03,080 really understand what it is that you're doing and the risks 1520 01:28:03,080 --> 01:28:07,320 that you're taking and have this beautiful family and, and, and, 1521 01:28:07,360 --> 01:28:09,640 and juggle the two things? Yeah. 1522 01:28:10,840 --> 01:28:15,560 Well, I'd, I'd start off by saying our profession, test 1523 01:28:15,560 --> 01:28:18,160 professional, test pilot and test engineers are part of that 1524 01:28:18,160 --> 01:28:20,920 and test sizzles and wizzles are all amazing. 1525 01:28:22,240 --> 01:28:27,200 It's really not about taking risk, it's about mitigating risk 1526 01:28:27,600 --> 01:28:32,800 and learning how to push the boundaries in a very deliberate, 1527 01:28:33,320 --> 01:28:38,600 methodical and safe fashion. With that being said, yes, we 1528 01:28:38,600 --> 01:28:44,120 are doing things in planes that are not always, that are not 1529 01:28:44,120 --> 01:28:47,840 typical, put it that way, that sometimes have never been done. 1530 01:28:48,480 --> 01:28:54,600 And we're, we're doing that because we and I have a passion 1531 01:28:54,960 --> 01:29:01,160 and a purpose to drive to try to ensure our war fighter is 1532 01:29:01,160 --> 01:29:03,880 prepared for whatever they may be called to do. 1533 01:29:03,880 --> 01:29:08,040 Which means they need systems, they need weapons and aircraft 1534 01:29:08,840 --> 01:29:11,560 that are going to work the way that they're supposed to work. 1535 01:29:11,840 --> 01:29:14,640 And we need people that are able to go out there and ensure that 1536 01:29:14,640 --> 01:29:20,800 that is going to happen. And I've always wanted to serve. 1537 01:29:21,840 --> 01:29:24,800 I never knew it was going to look like my parents were 1538 01:29:24,800 --> 01:29:26,680 firefighters. My mom was a paramedic on top of 1539 01:29:26,680 --> 01:29:28,440 that. And they served and they taught 1540 01:29:28,440 --> 01:29:31,360 me a lot through their service. And I didn't want, I mentioned 1541 01:29:31,360 --> 01:29:33,360 earlier, I never planned on joining the military, but I did 1542 01:29:33,360 --> 01:29:37,600 have this desire to serve. And so when I, right before 1543 01:29:37,600 --> 01:29:40,360 going to college, Saving Private Ryan came out and when I went 1544 01:29:40,360 --> 01:29:45,080 and saw that movie and it is very, it was very impactful. 1545 01:29:45,160 --> 01:29:47,480 And oh, by the way, like I love Braveheart was my favourite 1546 01:29:47,480 --> 01:29:50,240 movie for years. So it was also this idea of like 1547 01:29:50,240 --> 01:29:52,160 fighting for freedom. And the same Private Ryan comes 1548 01:29:52,160 --> 01:29:56,520 out and you're like, OK, you know, what we have here is to be 1549 01:29:56,520 --> 01:29:59,720 celebrated and is not free, right? 1550 01:29:59,840 --> 01:30:04,200 It took millions upon millions of people throughout our history 1551 01:30:04,200 --> 01:30:06,920 and Western civilization of people sacrificing. 1552 01:30:06,920 --> 01:30:09,720 And that could be for causes, yes, from World War 2. 1553 01:30:09,720 --> 01:30:13,280 But you go back and back, and not just in America, right, In 1554 01:30:13,280 --> 01:30:14,920 England and other countries, like people have been 1555 01:30:14,920 --> 01:30:19,880 sacrificing in, in a way that is really beautiful and meaningful. 1556 01:30:20,120 --> 01:30:24,040 And so for me, I knew I wanted to serve. 1557 01:30:24,040 --> 01:30:27,200 And as I started serving, like this seemed to be the natural 1558 01:30:27,200 --> 01:30:29,760 way to do it. And we have calls, right? 1559 01:30:29,760 --> 01:30:32,640 And, and the call that was on my heart was to serve in this 1560 01:30:32,640 --> 01:30:36,120 fashion, this capacity. And I think I would be doing a 1561 01:30:36,120 --> 01:30:43,040 disservice to myself and to my family if I didn't, if, if I 1562 01:30:43,040 --> 01:30:45,800 didn't go out the door and pursue something that was 1563 01:30:45,800 --> 01:30:49,360 greater than myself, that was more meaningful and not playing 1564 01:30:49,360 --> 01:30:52,520 it safe per SE, but living my life to the fullest. 1565 01:30:52,840 --> 01:30:57,440 And, and that is, I think where joy is found, where true peace 1566 01:30:57,440 --> 01:30:59,760 is found, where fulfilment is found. 1567 01:30:59,960 --> 01:31:04,440 And so like it, it, it was never a question of like, I'm going to 1568 01:31:04,440 --> 01:31:06,440 do this. Is it worth it for, you know, 1569 01:31:06,560 --> 01:31:09,360 the risk of my life? Like we, we need to be able to 1570 01:31:09,360 --> 01:31:11,760 live our lives. But that does play into my faith 1571 01:31:11,760 --> 01:31:13,880 a little bit too. Like, let me be clear. 1572 01:31:14,640 --> 01:31:17,800 And this book is not a, it's not a Christian book or anything. 1573 01:31:19,480 --> 01:31:21,000 My faith is really important to me. 1574 01:31:21,000 --> 01:31:23,080 And it is a Christian faith. And I found it later in life, 1575 01:31:24,280 --> 01:31:26,120 you know, but that's beside the point. 1576 01:31:26,240 --> 01:31:31,760 The point is talking about our spiritual pillar, like 1577 01:31:31,880 --> 01:31:35,600 connecting in deeper spiritually is really important. 1578 01:31:35,640 --> 01:31:39,160 And I do think we do a disservice and not talking about 1579 01:31:39,160 --> 01:31:42,160 it more. The problem becomes I don't 1580 01:31:42,160 --> 01:31:44,240 think a lot of people know how to talk about it without 1581 01:31:44,240 --> 01:31:47,480 proselytising, without like whatever they believe, 1582 01:31:47,480 --> 01:31:49,960 proselytising that belief. And just like connecting with 1583 01:31:49,960 --> 01:31:54,040 people and appreciating like people's journeys, like that's 1584 01:31:54,040 --> 01:31:55,960 their journey and that's what they're going on. 1585 01:31:56,320 --> 01:32:01,040 And why I bring it up even in this conversation is for me, I 1586 01:32:01,040 --> 01:32:03,880 have a belief in Providence that I think things happen for a 1587 01:32:03,880 --> 01:32:06,200 reason. I don't know if I know what 1588 01:32:06,200 --> 01:32:09,080 those reasons are. And that reason may be me dying, 1589 01:32:09,400 --> 01:32:11,080 right? Like I may go through something 1590 01:32:11,080 --> 01:32:13,960 and the outcome is me like not making it. 1591 01:32:14,080 --> 01:32:17,120 But I think that there's still a grander purpose behind that. 1592 01:32:17,320 --> 01:32:21,840 So maybe that's, you know, through my death, you know, 1593 01:32:21,920 --> 01:32:27,440 someone else's journey is being formed and right, like, I don't 1594 01:32:27,440 --> 01:32:29,280 know where that line is. I'm not sure. 1595 01:32:29,280 --> 01:32:32,080 Like I'm not God, right? Like I'm just, I'm, I'm here to 1596 01:32:32,080 --> 01:32:34,280 serve and, and living my life in that fashion. 1597 01:32:34,280 --> 01:32:38,920 So that also makes my risk calculus or my approach to like, 1598 01:32:38,920 --> 01:32:42,120 what should I be doing? Like really centred on something 1599 01:32:42,160 --> 01:32:45,320 much bigger than myself. That's very interesting because 1600 01:32:45,320 --> 01:32:48,400 I think that so I would absolutely agree with you. 1601 01:32:48,400 --> 01:32:51,320 The book is not a, a preachy book. 1602 01:32:51,320 --> 01:32:53,360 That's not, you're not a proponent for Christianity in 1603 01:32:53,360 --> 01:32:55,160 that sense. You're not trying to convert 1604 01:32:55,160 --> 01:32:57,800 people to, to, to your, your, your faith. 1605 01:32:58,120 --> 01:32:59,640 So that that should be absolutely clear. 1606 01:33:00,120 --> 01:33:04,240 I mean, the, the, the book has warmth and it has humility and 1607 01:33:04,240 --> 01:33:07,120 yourself deprecating and self effacing. 1608 01:33:07,120 --> 01:33:10,400 I think those are qualities you don't normally see from fighter 1609 01:33:10,400 --> 01:33:12,320 pilots. You know, you, you put your 1610 01:33:12,320 --> 01:33:14,640 hands up and say, I screwed this up and this is the lesson I 1611 01:33:14,640 --> 01:33:17,040 learned from it. And that's the theme through the 1612 01:33:17,040 --> 01:33:19,520 book and the things that you didn't do well and, and the fact 1613 01:33:19,520 --> 01:33:23,080 that then even when you succeed, your failures can still be on 1614 01:33:23,080 --> 01:33:25,720 your mind. You talked about that aim, that 1615 01:33:25,920 --> 01:33:28,560 shooting the aim 9 from the F35 and the pickle button. 1616 01:33:28,560 --> 01:33:32,080 And that's a so I would, I would absolutely recommend that people 1617 01:33:32,080 --> 01:33:35,680 buy the book and, and read it, even if, I mean, if, if you just 1618 01:33:35,680 --> 01:33:37,560 want call aviation stories, they're in there. 1619 01:33:37,800 --> 01:33:39,760 So it's a, it's a mix of both those things. 1620 01:33:40,360 --> 01:33:43,840 Tell, tell us a little bit about how people can get hold of it. 1621 01:33:43,840 --> 01:33:45,520 Is it going to be on Amazon? At the moment? 1622 01:33:45,520 --> 01:33:48,160 I've read a pre pre release copy, digital copy. 1623 01:33:48,160 --> 01:33:50,440 I haven't got a physical copy but can they go and buy it on 1624 01:33:50,440 --> 01:33:51,760 Amazon and will it be in the bookstores? 1625 01:33:52,680 --> 01:33:55,080 Yeah, no thank you. You can go to my website, 1626 01:33:55,080 --> 01:33:57,400 tuckerhamilton.com. It's super easy, Tucker 1627 01:33:57,400 --> 01:34:03,080 hamilton.com and there's ways that you can click on that links 1628 01:34:03,080 --> 01:34:06,160 to go buy the book. You can find them. 1629 01:34:06,160 --> 01:34:11,280 Amazon UK has it. Amazon and America has it. 1630 01:34:11,800 --> 01:34:13,960 Barnes and Noble. Yeah, it's basically where 1631 01:34:13,960 --> 01:34:19,840 wherever you usually get books, Audible has audio version that 1632 01:34:19,840 --> 01:34:22,480 is also publishing on the 24th of March. 1633 01:34:22,920 --> 01:34:26,040 And yeah, I'd love to hear from people like your thoughts, 1634 01:34:26,520 --> 01:34:28,960 listeners like what they think, if they have other questions. 1635 01:34:29,080 --> 01:34:32,080 You can reach me on Tucker hamilton.com and just ask me 1636 01:34:32,080 --> 01:34:34,000 questions. I try to be very open, 1637 01:34:34,000 --> 01:34:38,400 transparent about like my life and my perspective. 1638 01:34:38,400 --> 01:34:41,640 I just, I will always caveat that it's a biassed perspective, 1639 01:34:42,000 --> 01:34:44,160 right? Like it's mine and it's 1640 01:34:44,160 --> 01:34:49,360 definitely not for everyone. Or I'm like my, my truth that I 1641 01:34:49,360 --> 01:34:53,320 may imagine is right for someone may not be their truth right 1642 01:34:53,320 --> 01:34:56,520 like that. Like like I said, we all have 1643 01:34:56,520 --> 01:34:58,920 our own journey, but I'm happy to to share in. 1644 01:34:59,560 --> 01:35:02,000 Yeah, I'd love for people to to tell me what they think. 1645 01:35:02,000 --> 01:35:05,200 And I really appreciate you allowing me to come share. 1646 01:35:05,560 --> 01:35:10,000 You know some of my stories. Thank you so much for for being 1647 01:35:10,000 --> 01:35:11,720 on the channel and it's been a pleasure speaking to you. 1648 01:35:11,720 --> 01:35:15,080 I do hope that when you're not too busy on fronting the the 1649 01:35:15,080 --> 01:35:18,560 Fire podcast, you'll come back on and talk about MC 12 standing 1650 01:35:18,560 --> 01:35:22,040 up and being a test pilot and flying a 35 and being 1651 01:35:22,400 --> 01:35:24,120 responsible for a 35 test at Edwards. 1652 01:35:24,120 --> 01:35:26,720 That sounds amazing. So hopefully you'll come back on 1653 01:35:26,720 --> 01:35:28,840 and do that. But I wish you the very best of 1654 01:35:28,840 --> 01:35:29,760 luck with the book. It is. 1655 01:35:29,760 --> 01:35:34,320 It's a really good. Oh no, I I was gonna please tell 1656 01:35:34,320 --> 01:35:38,080 me more about positive things. I was gonna interrupt and just 1657 01:35:38,080 --> 01:35:42,120 say I should really quick and I have to run here in a second as 1658 01:35:42,120 --> 01:35:43,880 you do as well. But I should really quick tell 1659 01:35:43,880 --> 01:35:46,560 you my call sign story. Oh, I thought you were gonna 1660 01:35:46,560 --> 01:35:48,960 make me buy you because you said you're gonna drink, but I. 1661 01:35:49,000 --> 01:35:51,040 No, I could do it. I could do it on. 1662 01:35:51,240 --> 01:35:52,920 I'll do it for your. Where does it come from? 1663 01:35:53,360 --> 01:35:55,920 Yeah, yeah. So it is in the book. 1664 01:35:55,920 --> 01:35:57,680 So you can go. You can go by and read the book 1665 01:35:57,680 --> 01:36:01,800 too. So there I was and we get call 1666 01:36:01,800 --> 01:36:07,160 signs by either like a natural or a natural is like I have a 1667 01:36:07,160 --> 01:36:10,240 buddy, this last name is Stevens and so his call sign is Even 1668 01:36:10,320 --> 01:36:12,720 Stevens, even Stevens. Or I have a buddy's last name is 1669 01:36:12,880 --> 01:36:17,320 Abel and his call sign is Kanan. So Cain and Abel or you know 1670 01:36:17,320 --> 01:36:21,240 like naturals or you do something stupid and mine's the 1671 01:36:21,240 --> 01:36:23,360 latter. That is the typical of what you 1672 01:36:23,560 --> 01:36:26,080 get from people. So there I was, brand new 1673 01:36:26,080 --> 01:36:27,720 wingman. I was flying on the wing of my 1674 01:36:27,720 --> 01:36:31,080 flight lead and we are doing air to air combat training. 1675 01:36:31,280 --> 01:36:34,200 So we're going out the door and there are two of us against, you 1676 01:36:34,200 --> 01:36:37,360 know, a number of bad guys. And we're, we're supposed to 1677 01:36:37,360 --> 01:36:41,080 like get the bad guys out and, you know, kill all the bad guys 1678 01:36:41,080 --> 01:36:42,960 out in front of us and then go to a target area, protect the 1679 01:36:42,960 --> 01:36:44,440 target. The bad guys are just other 1680 01:36:44,600 --> 01:36:46,920 people from our squadron that fly against us and emulate being 1681 01:36:46,920 --> 01:36:49,800 a threat either way. So we shoot our missiles at the 1682 01:36:49,800 --> 01:36:53,720 bad guys and we turn sideways and we wait for our missiles to 1683 01:36:53,720 --> 01:36:56,600 do their work. They're Amrams, so you know, 1684 01:36:56,600 --> 01:37:00,240 they, they're able to go finish the job as we're sideways and 1685 01:37:00,240 --> 01:37:02,080 then we turn back into the target area. 1686 01:37:02,920 --> 01:37:05,880 And if people aren't watching this online, I will describe 1687 01:37:06,480 --> 01:37:09,600 this, this position or this manoeuvre. 1688 01:37:09,600 --> 01:37:12,360 Part of me puts me in this situation because I'm this 1689 01:37:12,360 --> 01:37:16,000 aircraft, it puts me out of position, but I still see my 1690 01:37:16,000 --> 01:37:18,160 flight lead. And so my flight leads like 2 1691 01:37:18,560 --> 01:37:20,200 one's blind. I don't see you, which just 1692 01:37:20,200 --> 01:37:21,720 means like blind, like I don't see you. 1693 01:37:22,640 --> 01:37:26,360 OK, number one, I'm at your left 7:00, five miles, which just 1694 01:37:26,360 --> 01:37:29,080 means I'm at the, you know, clock position. 12:00 is right 1695 01:37:29,080 --> 01:37:32,040 off your nose, 6:00 behind you left. 7:00, five miles, which 1696 01:37:32,040 --> 01:37:33,480 means I'm out of position. But I still see him. 1697 01:37:33,480 --> 01:37:35,560 So he has some assurances that I'm, I'm getting back in 1698 01:37:35,560 --> 01:37:37,280 position. So I start getting back into 1699 01:37:37,280 --> 01:37:41,240 position. 30 seconds goes by and he's like, you know, 2. 1700 01:37:41,240 --> 01:37:43,360 I still don't see where you at. I'm like one, I'm I'm right of 1701 01:37:43,360 --> 01:37:45,920 your left wing now. 9:00 five miles. 1702 01:37:45,920 --> 01:37:48,920 I'll bring it in because we usually fly closer than 5 miles, 1703 01:37:50,320 --> 01:37:52,320 but we don't fly that close. It's still a couple miles away. 1704 01:37:52,320 --> 01:37:54,560 It's still hard to see each other for sure. 5 miles is 1705 01:37:54,560 --> 01:37:56,720 really hard to see someone else. But I'm, I'm, I'm seeing him. 1706 01:37:57,400 --> 01:38:00,240 He's like all right, 30 seconds goes by two, I still don't see 1707 01:38:00,240 --> 01:38:02,120 you. OK, one, I'm at your left, nine 1708 01:38:02,120 --> 01:38:04,960 O clock in position now, right where I'm supposed to be. 1709 01:38:05,320 --> 01:38:07,160 Still doesn't see me. So I start moving closer. 1710 01:38:07,360 --> 01:38:09,840 He's like 2-2. Where are you? 1711 01:38:09,840 --> 01:38:13,600 I still don't see you, man. One, I'm at your left, you know, 1712 01:38:13,600 --> 01:38:15,640 9 O clock now one mile with a flash. 1713 01:38:15,640 --> 01:38:18,720 So I start rocking my wings back and forth like it's hard to see 1714 01:38:19,120 --> 01:38:21,480 jet that isn't rocking the wings, but easy to see as you 1715 01:38:21,480 --> 01:38:24,560 flash your wing. He's he doesn't see me still. 1716 01:38:24,560 --> 01:38:26,120 I'm getting upset. He's getting upset. 1717 01:38:26,120 --> 01:38:29,240 And I'm like, I'm right here. I'm rocking my wing. 30 seconds 1718 01:38:29,240 --> 01:38:30,440 goes by. He's like 2. 1719 01:38:30,760 --> 01:38:33,320 Where are you? And I'm getting close and I'm 1720 01:38:33,320 --> 01:38:36,040 like 3000 feet and I'm like one, I'm, I'm right off your left 1721 01:38:36,040 --> 01:38:38,560 wing. I'm, I'm rocking my wings and 1722 01:38:38,560 --> 01:38:40,440 I'm flaring. I'm start dropping flares behind 1723 01:38:40,440 --> 01:38:44,560 me like fireworks. There's a pause on the radio. 2, 1724 01:38:44,920 --> 01:38:47,440 Are you flying off one of the bad guys we killed out? 1725 01:38:48,200 --> 01:38:50,560 And I looked down at the screen that I had in front of me the 1726 01:38:50,560 --> 01:38:52,960 entire time. This screen was in front of me 1727 01:38:53,800 --> 01:38:56,000 and I am 60 miles away from my flight lead. 1728 01:38:56,480 --> 01:39:00,600 Because what I did was when we turned sideways to shooter 1729 01:39:00,600 --> 01:39:04,000 missiles, I saw one of the bad guys that we killed out turned 1730 01:39:04,000 --> 01:39:07,560 southbound basically, and I just flew off of his wing. 1731 01:39:07,560 --> 01:39:09,960 This entire time I'm flying off the wrong person's wing. 1732 01:39:10,120 --> 01:39:13,200 My flight lead is all alone. He's getting close to the target 1733 01:39:13,200 --> 01:39:17,480 area and he's like 2 get back in position, you know, like just 1734 01:39:17,480 --> 01:39:20,560 furious. And I'm like what the worst 1735 01:39:20,560 --> 01:39:24,520 pilot ever and I plug it into 5 stages of afterburner to race 1736 01:39:24,520 --> 01:39:28,440 back into position to get on onto his wing and I get there 1737 01:39:28,600 --> 01:39:31,920 Afterburner is amazing. But the the problem with 1738 01:39:31,920 --> 01:39:34,720 afterburner is you use all your fuel when you do that. 1739 01:39:34,720 --> 01:39:37,160 So I rejoined on his wing and we're supposed to be there for 1740 01:39:37,160 --> 01:39:40,080 like 15 minutes and I am out of fuel within 10 seconds. 1741 01:39:40,080 --> 01:39:43,560 I don't even do a complete turn. My bingo light goes on and says 1742 01:39:43,560 --> 01:39:46,840 bingo fuel, which means I have to immediately return to base. 1743 01:39:47,040 --> 01:39:48,600 And he's just like, get out of here. 1744 01:39:48,600 --> 01:39:53,680 Like just totally pissed. But so many lessons, like 1, I 1745 01:39:53,680 --> 01:39:56,520 made this mistake. I got channelized like looking 1746 01:39:56,520 --> 01:39:59,680 at him that looking at the wrong aircraft the entire time instead 1747 01:39:59,680 --> 01:40:01,800 of just looking down at a scope that I had in front of me. 1748 01:40:02,160 --> 01:40:04,360 And I'm getting upset with him when I was really the one that 1749 01:40:04,360 --> 01:40:06,120 was, you know, and we do that often, right? 1750 01:40:06,120 --> 01:40:09,680 Like we, we cause problems or problems happen to us as well. 1751 01:40:09,880 --> 01:40:12,240 And we find ourselves in a situation where I'm like, and 1752 01:40:12,680 --> 01:40:15,520 what did I do? Like the completely wrong thing 1753 01:40:15,520 --> 01:40:18,680 to do, which was use all my fuel to get back in position where I 1754 01:40:18,680 --> 01:40:20,560 should have just, you know, gone fast. 1755 01:40:20,680 --> 01:40:24,840 But not gone like afterburner fast to get back in position. 1756 01:40:24,840 --> 01:40:27,280 I would have had enough fuel then to continue the mission. 1757 01:40:27,640 --> 01:40:30,560 And so they, they gave me the call sign Cinco. 1758 01:40:30,680 --> 01:40:35,840 So Uno, Uno dos tres cuatro Cinco 55 stages of afterburner 1759 01:40:35,840 --> 01:40:38,280 is as fast as you can go in the F15C. 1760 01:40:38,560 --> 01:40:41,480 And they, they gave me Cinco for flying around in five stages of 1761 01:40:41,480 --> 01:40:46,400 AB for too long. So there's my, my stupid, stupid 1762 01:40:46,440 --> 01:40:48,720 Cinco story. That's brilliant. 1763 01:40:50,480 --> 01:40:51,600 So good. Thank you so much for sharing 1764 01:40:51,600 --> 01:40:52,920 that with us. Thank you for sharing your 1765 01:40:52,920 --> 01:40:54,280 thoughts and your experiences on AI. 1766 01:40:54,280 --> 01:40:56,560 Thank you for bringing the book to us. 1767 01:40:56,560 --> 01:40:58,680 I hope people buy it. I hope they do feedback to you. 1768 01:40:58,680 --> 01:41:01,400 It's brilliant to read. I wish you the best of luck with 1769 01:41:01,400 --> 01:41:04,920 it and good luck with the final pilot podcast and come back on 1770 01:41:04,920 --> 01:41:06,640 the channel and see me again when you when you've got time. 1771 01:41:07,160 --> 01:41:08,560 Yeah, fantastic. Thank you so much. 1772 01:41:08,760 --> 01:41:09,800 All right, all my best.