YouTube2Text

Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Humanoids | EP #295 — Transcript

by Peter H. Diamandis · 7,244 words · 1,122 segments · language en · Watch on YouTube

Full transcript

  1. 0:05Wow, it's so much brighter out here.
  2. 0:07>> It is. It's, you know, the future is
  3. 0:09bright.
  4. 0:10>> There's lens flare in the future.
  5. 0:12>> Yeah.
  6. 0:13So, welcome everybody to the Oscars of
  7. 0:15optimism. That, of course, we're opening
  8. 0:18up the summit with the king of
  9. 0:20AUTONOMOUS WEAPON SYSTEMS.
  10. 0:22>> [laughter]
  11. 0:26>> ALL RIGHT. SO,
  12. 0:29I'M GOING TO start with the fact that
  13. 0:31we're awarding three X prizes this week.
  14. 0:34Palmer, congratulations for winning our
  15. 0:37wildfire X prize.
  16. 0:39>> Thank you so much.
  17. 0:41>> [applause]
  18. 0:42>> This one's been a long time in in the
  19. 0:44works.
  20. 0:45>> Let's get Let's get the the the slide up
  21. 0:47here. So, Palmer won $2.2 million for
  22. 0:50Anduril.
  23. 0:51>> As if you needed it.
  24. 0:52>> Um a million of
  25. 0:54a million of that from Lockheed,
  26. 0:57which was kind of interesting.
  27. 1:00>> We do a lot of work with Lockheed. We
  28. 1:01also compete with them, but we
  29. 1:04in the in the defense space, it's you
  30. 1:06know, we we we compete with their
  31. 1:07fighter jet division and we partner with
  32. 1:09them on some of their payloads. And so,
  33. 1:11we do a lot of stuff together. But, you
  34. 1:12me and you have both been trying to get
  35. 1:13this autonomous firefighting stuff going
  36. 1:15for a long time.
  37. 1:16>> mean, you got this prize announced back
  38. 1:18in 2019. It's been a long time.
  39. 1:20Anduril's one of our first two products
  40. 1:22was an autonomous firefighting tank that
  41. 1:25we started building in 2017, finished in
  42. 1:272019.
  43. 1:28>> It's insane that we let fires become
  44. 1:30conflagrations before we put them out.
  45. 1:32>> And it well, and it's and it's kind of
  46. 1:33insane that people are saying, "Oh, the
  47. 1:36you know, the technology's finally made
  48. 1:37it possible." And I don't know, I'm not
  49. 1:39going to correct them too often in
  50. 1:41public, but it's I mean, this has been
  51. 1:42possible for over a decade. It's
  52. 1:44>> Yeah. All right.
  53. 1:44>> It's just that
  54. 1:45it's become politically politically
  55. 1:47impossible to ignore the problem now.
  56. 1:49>> We'll watch a 1-minute video clip of
  57. 1:51what your team achieved. And if you walk
  58. 1:53us through it here.
  59. 1:56>> Sure. So, you
  60. 1:57this was
  61. 1:59an example of what we were doing at the
  62. 2:00test site where we were
  63. 2:02detecting fires, classifying fires. What
  64. 2:05you don't really see in here is that you
  65. 2:06also need [music] to reject false fires.
  66. 2:08That was part of the test as well. So,
  67. 2:10you have to know, you know, who's the
  68. 2:11guy smoking a cigarette? Who's the guy
  69. 2:13What's What's What's a campfire? What's
  70. 2:14a road flare? And then what's something
  71. 2:16that shows the characteristics of real
  72. 2:17fire? That's what you need to flag to
  73. 2:20people. That's what you need to flag to
  74. 2:22emergency services. Uh and then you need
  75. 2:24to match make the type of fire to the
  76. 2:27right asset that's in the area. And so,
  77. 2:29you want to basically use the most
  78. 2:31cost-effective thing that can respond
  79. 2:34and get there in time. And if
  80. 2:36something's growing faster, you might
  81. 2:37need something different. If it's in
  82. 2:39woods versus on a brush hill, you might
  83. 2:41need something different. And so, you
  84. 2:42know, it's it's about detection,
  85. 2:44classification, rejection, matchmaking,
  86. 2:47and then going and and hitting the
  87. 2:48target. And the last thing I'll say is
  88. 2:49that with most of what we do,
  89. 2:52uh our work with the military,
  90. 2:55you'd go through many of these same
  91. 2:56steps,
  92. 2:57but if you find a target, let's say it
  93. 3:00takes you, you know, an hour to an hour
  94. 3:03to to get to somebody versus 10 minutes,
  95. 3:05it usually doesn't make as much of a
  96. 3:08difference. When you're talking about
  97. 3:10fire, though, the difference between 10
  98. 3:12minutes and an hour response, totally
  99. 3:15different tools you're going to need to
  100. 3:16respond with. You know,
  101. 3:18a bomb is is is a is a bomb whether or
  102. 3:20not it takes a day or an hour or a
  103. 3:22minute to get there. Uh but when you
  104. 3:24when when a fire is is is is
  105. 3:25metastasizing, you need to pick the
  106. 3:27right thing that can put it out before
  107. 3:29it gets too large.
  108. 3:30>> Amazing.
  109. 3:31>> You know, uh
  110. 3:32we're going to be announcing something
  111. 3:33in a month's time
  112. 3:35uh that you're funding a new X Prize,
  113. 3:39and uh it's going to be secret until
  114. 3:41then, but we're going to tease it now.
  115. 3:44Uh
  116. 3:45another $10 million prize. Would you
  117. 3:47share share with us what we're going to
  118. 3:49do?
  119. 3:50>> Sure, you know, uh
  120. 3:52the
  121. 3:53you know, we we should we should keep
  122. 3:54the details secret cuz there's there's
  123. 3:56some there's some good twists, but
  124. 3:58suffice to say that I'm not I'm not
  125. 4:01interested in joining the race to push
  126. 4:04AI forward with my money. I think that
  127. 4:07the
  128. 4:08the artificial intelligence side of
  129. 4:10models is is is pretty well covered by
  130. 4:12plenty of people. I think we've not been
  131. 4:14doing enough understanding biological
  132. 4:16intelligence particularly beyond people
  133. 4:18and so this is going to be a prize
  134. 4:20around understanding how animals think
  135. 4:23and making so that we can communicate
  136. 4:25with them 10 times better than the
  137. 4:27current state of the art for the current
  138. 4:29smartest species and not just trained
  139. 4:30animals, but wild animals in their own
  140. 4:33in their own language.
  141. 4:34>> Amazing. [applause and cheering]
  142. 4:36>> And I think that there's actually a lot
  143. 4:37that we're going to be able to learn. I
  144. 4:38mean, if you look at large language
  145. 4:40models, they're kind of this, you know,
  146. 4:42they're they're they're based on
  147. 4:44humanity's way of thinking and
  148. 4:46communicating distilled and I think that
  149. 4:48there's plenty to learn from all of the
  150. 4:51other species that are on this planet
  151. 4:53and I wouldn't be surprised if it has
  152. 4:54applications for AI particularly AI
  153. 4:58where you need to be
  154. 4:59more efficient, you know, more power
  155. 5:01efficient, more time efficient,
  156. 5:03you know,
  157. 5:04a different point of view, two heads are
  158. 5:06better than one, whichever whatever
  159. 5:09metaphor you want to use.
  160. 5:10>> Love it. Love it. All right, I'm going
  161. 5:11to kick off the first official question
  162. 5:13here.
  163. 5:14P-doom, you're in the AI industry, the
  164. 5:16defense industry, what's your P-doom?
  165. 5:20>> I'm just not worried
  166. 5:23about AI dooming
  167. 5:25the I guess I'm not worried about AI
  168. 5:28doing the killing. I'm very very worried
  169. 5:30about evil people
  170. 5:31>> Yes.
  171. 5:32>> using moderately competent AI to do evil
  172. 5:35things.
  173. 5:35>> Exactly.
  174. 5:36>> And I am s-
  175. 5:39kind of frustrated by a lot of people
  176. 5:41who are new to the dooming hobby.
  177. 5:45It's It's It's sort It's sort of like
  178. 5:48how we had to deal with
  179. 5:49the war in Ukraine happens and all of a
  180. 5:51sudden all these people who have had the
  181. 5:53dumbest opinions you've ever heard about
  182. 5:55the military for their whole life
  183. 5:58and people who believed we lived at the
  184. 5:59end of history that we had moved past
  185. 6:02conflict that we lived in a world where
  186. 6:04you could just write a nasty letter in
  187. 6:06the UN forgetting that that only works
  188. 6:08if it's backed up by credible threat of
  189. 6:10violence and then all of a sudden they
  190. 6:11change their minds. You're seeing this
  191. 6:12with AI. People who have not thought
  192. 6:15about this problem for more than
  193. 6:17a couple years or maybe even just a
  194. 6:19couple months
  195. 6:21are saying oh my god this is the biggest
  196. 6:23threat to humanity. These are people who
  197. 6:25have spent almost no time actually
  198. 6:27analyzing what the largest threats to
  199. 6:29humanity are. They haven't spoken to any
  200. 6:31of the experts in this area. They
  201. 6:33certainly weren't buying freeze-dried
  202. 6:34food and building prepper bunkers, you
  203. 6:37know, like like like like me and many of
  204. 6:39my my my buddies who who have who have
  205. 6:42indulged fantasies about what doom could
  206. 6:44look like and also thought realistically
  207. 6:46about it. And so yeah, the AI doom is
  208. 6:49way down the list for me. I'm I'm I'm
  209. 6:51I'm a big optimist. I think it's going
  210. 6:52to massively reduce the cost of resource
  211. 6:54extraction, processing, transformation.
  212. 6:57You're going to see every part of
  213. 6:58society become subject to the same
  214. 7:01economies of scale and efficiency that
  215. 7:03automated industries that automated
  216. 7:05without AI have seen. So think like
  217. 7:08agriculture or or textiles. You can buy
  218. 7:122 weeks of calories for less than 1 hour
  219. 7:17of federal minimum wage labor. That's
  220. 7:19crazy.
  221. 7:20>> Let's give it up for
  222. 7:21Palmer here. You're going to see that
  223. 7:23for everything.
  224. 7:24>> David [applause]
  225. 7:26>> So I I got a really
  226. 7:27um
  227. 7:28pressing question on my mind. I'm kind
  228. 7:29of a feeds and feeds and ships and AI
  229. 7:31foundation models guy.
  230. 7:33And so when you're putting systems in
  231. 7:35the field, weapons in the field, uh
  232. 7:37you can you can put the intelligence
  233. 7:39inside it and it this has always been a
  234. 7:41big problem for the Air Force, like if
  235. 7:42we're going to start flying drones
  236. 7:43instead of pilots, who's going to
  237. 7:45unleash the weapons and how's that
  238. 7:46decision going to get made? And the the
  239. 7:48downside of transmitting it back to
  240. 7:50headquarters is you have latency.
  241. 7:53Uh the upside is you have a lot more
  242. 7:54compute. Uh what you can put locally in
  243. 7:57a device, in a robot in the field, in a
  244. 7:58plane is limited, but it's really really
  245. 8:01fast. So what's your philosophy on like,
  246. 8:03you know, I think a lot of people would
  247. 8:04say, well, are we going to have
  248. 8:06self-contained decision-making weapons
  249. 8:08wandering around enemy territories or
  250. 8:10are we going to have like are we going
  251. 8:11to have you know, control at the
  252. 8:13headquarters?
  253. 8:13>> Yeah, more or less. The the real concern
  254. 8:15isn't so much the latency or or the
  255. 8:19speed of computation. The problem is
  256. 8:21that if you rely on
  257. 8:24you know, a central a central location
  258. 8:25to do your compute, you now need to have
  259. 8:28command and control that has a good
  260. 8:31high-bandwidth communication back and
  261. 8:33forth between you and potentially
  262. 8:35thousands or tens of thousands of active
  263. 8:37assets. And if your enemy jams your
  264. 8:39communications or intercepts your
  265. 8:41communications or hacks your
  266. 8:42communications, you're done.
  267. 8:44Uh or worse, they just skip targeting
  268. 8:47all the stuff in the field and they go
  269. 8:49for the place where it's all being
  270. 8:51controlled. You know, we saw we saw this
  271. 8:53I there was a there was a there there
  272. 8:55was there was this documentary that
  273. 8:57George Lucas was working on and and the
  274. 8:59Trade Federation, they had they had
  275. 9:01their central AI command and control
  276. 9:03blown up and that was how the Battle of
  277. 9:05Naboo actually resolved was all all the
  278. 9:08battle droids lost
  279. 9:09>> Oh yeah, they all fell over.
  280. 9:10>> That's right, they all fell over. And so
  281. 9:12I think that that's that that's the
  282. 9:13practical reality that's driving the
  283. 9:15need to have autonomy living not back
  284. 9:18home, but forward. There's other areas
  285. 9:20like you go underwater, you can't talk
  286. 9:23to things. And sometimes you could talk
  287. 9:25to things, but you don't want to. Having
  288. 9:27radio emissions actually exposes you
  289. 9:29necessarily. And so what what you
  290. 9:32for better or for worse, you're going to
  291. 9:33need this. I actually don't think it's
  292. 9:34as much of a shift as people think.
  293. 9:37I there's a lot of people who look at
  294. 9:38this and say, "Oh my god, Pandora's Box
  295. 9:40is being opened. You're going to have
  296. 9:41weapons that are making their own kill
  297. 9:43decisions." And they don't recognize the
  298. 9:46reality, which is that that's been the
  299. 9:47case for decades. You look at the You
  300. 9:49look at the You look at the Aegis system
  301. 9:51on destroyers. It has a fully autonomous
  302. 9:53mode. You can flip a button and it will
  303. 9:55shoot down anything that comes within
  304. 9:57whatever range you programmed it with
  305. 10:00without any human confirmation. You look
  306. 10:02at SeaWiz and SeaRAM for shooting down
  307. 10:05incoming missiles, incoming incoming
  308. 10:07artillery fire, incoming missiles, or
  309. 10:10incoming speedboats.
  310. 10:11Same thing. You can put them in fully
  311. 10:13autonomous modes. And I mean, it Going
  312. 10:15back to into Vietnam, we had missiles
  313. 10:17that you could launch from an aircraft.
  314. 10:18They flew over the horizon, tried to
  315. 10:20identify the right radio emitters that
  316. 10:22they believed corresponded with a
  317. 10:24surface-to-air missile launch site, and
  318. 10:26then struck them without any The person
  319. 10:28had to decide, "I'm going to send it in
  320. 10:29this area. They're going to look for
  321. 10:31this signal, and I know that there's
  322. 10:32this probability they're going to get it
  323. 10:34right." That decision is going to remain
  324. 10:36with people. But, autonomous weapons are
  325. 10:39widely widely There's a whole other talk
  326. 10:42we're not going to have today, but I
  327. 10:43will I I do have a rant about this where
  328. 10:46actually autonomous weapons have existed
  329. 10:47for thousands of years. If you look at
  330. 10:50Defined as kind of weapons that can
  331. 10:53execute the intent of what you
  332. 10:55programmed to do independent from the
  333. 10:57operator to the engineer. You look at,
  334. 10:59for example, pitfall traps that allow
  335. 11:01people to walk over them, but then
  336. 11:02capture the commander on his horse when
  337. 11:04he walks by and it's heavier. You look
  338. 11:06at certain poison traps that would that
  339. 11:09would kill people who had who had
  340. 11:12certain characteristics, not kill people
  341. 11:14who had other characteristics, or not
  342. 11:15kill animals.
  343. 11:17You look
  344. 11:18many types of landmines and sea mines
  345. 11:21and even even torpedoes. And so,
  346. 11:23it depends on on your exact definition.
  347. 11:26But, even going into myth where people
  348. 11:27are creating golems that can you go out
  349. 11:29and fight on their fight on their behalf
  350. 11:31or or defend you defend their city.
  351. 11:33I I guess my argument is we've had
  352. 11:35autonomous weapons for a very very long
  353. 11:37time. It's just a new set of technology
  354. 11:40that's that's making them better.
  355. 11:41>> So, technical question on that. The
  356. 11:42foundation models that go into those
  357. 11:44those remote systems
  358. 11:46>> Or the pit traps.
  359. 11:47>> Yeah, well so that that
  360. 11:49right right now the open source models
  361. 11:51are all Chinese. So, I'm pretty sure
  362. 11:53you're not going to be putting Chinese
  363. 11:54models into remote weapons. Are you
  364. 11:56going to build Anduril foundation models
  365. 11:58or how how are you going to get AI in
  366. 12:00those devices?
  367. 12:01>> I mean, we've been doing it since 2017.
  368. 12:03So, we're way ahead of everybody. We're
  369. 12:05we're a decade beyond that question.
  370. 12:08Um
  371. 12:09I think that we also don't the
  372. 12:11foundation models that people are
  373. 12:12building are not the types of models you
  374. 12:14want to be doing this. They're non that
  375. 12:16they're largely non-deterministic.
  376. 12:17They're non-auditable. It it's it's a
  377. 12:19very different category. The thing that
  378. 12:21we are building, unfortunately, has very
  379. 12:23few consumer applications.
  380. 12:25>> Mhm.
  381. 12:25>> And vice versa.
  382. 12:27Um there of course are LLM use cases for
  383. 12:30defense that make a lot of sense, but
  384. 12:31it's not guiding my missile or or
  385. 12:34classifying radar your radar radar radar
  386. 12:36feeds. I mean, and we we're also we
  387. 12:38can't build models that ingest text and
  388. 12:40then spit out text. I need to process
  389. 12:43raw data radar data feeds. I need to
  390. 12:45process sonar data from passive
  391. 12:47listening stations on submarines. That's
  392. 12:49it's it's pretty highly specialized. And
  393. 12:51I I will take some credit here.
  394. 12:53Uh all these people who are talking
  395. 12:55about how AI is going to change
  396. 12:56everything
  397. 12:57they are mostly uh
  398. 12:59recent converts.
  399. 13:01Anduril Industries, you might notice the
  400. 13:03acronym is AI. Um so, and that that was
  401. 13:06not that was not a mistake.
  402. 13:07>> Okay.
  403. 13:08>> Alex, jump in.
  404. 13:09>> Yeah,
  405. 13:10I think maybe dovetailing on our
  406. 13:12discussion of driving bench, we're now
  407. 13:14in an era when it's possible to take
  408. 13:16frontier models, drop them into a
  409. 13:17robotic embodiment, and get pretty
  410. 13:20stunning results. So, I'd love to ask
  411. 13:21you a question maybe about the scaling
  412. 13:23hypothesis as it applies to autonomous
  413. 13:26weapon systems or autonomous systems in
  414. 13:28general. If you have to extrapolate out,
  415. 13:30is there a Moore's law? Is there an
  416. 13:32experience curve for what as obviously
  417. 13:36the capabilities of underlying frontier
  418. 13:37models, whether they're from Frontier
  419. 13:39Labs proper or from AI,
  420. 13:42as as those progress,
  421. 13:44do we see any equivalent of a Moore's
  422. 13:47law for the autonomous systems
  423. 13:49themselves? And and maybe the the second
  424. 13:52half of the question would be, do you
  425. 13:54think ultimately we achieve some sort of
  426. 13:56convergency? Anduril it makes a variety
  427. 13:59of embodiments in a variety of different
  428. 14:01form factors. Do you see ultimately some
  429. 14:04general purpose category killer robot
  430. 14:07that is equivalently the generalist
  431. 14:10robotic embodiment just like we now have
  432. 14:12generalist models that are capable of
  433. 14:14doing substantially any task?
  434. 14:16>> No, I don't think so on the defense
  435. 14:18side. I think that for manufacturing and
  436. 14:21for hospitality and medicine, you might
  437. 14:23see this because you get economies of
  438. 14:25scale that uh that that that that that
  439. 14:27are really that are really material for
  440. 14:29those types of businesses and you know,
  441. 14:30a humanoid robot may not be the optimal
  442. 14:33form for a burger griller and it may not
  443. 14:35be the optimal form for a firefighter.
  444. 14:37It may not be the optimal form for a for
  445. 14:38a nurse [laughter] or an assembly
  446. 14:40technician.
  447. 14:41>> I see you're smiling.
  448. 14:42>> But even if it's not optimal, the fact
  449. 14:45that it is shared with humanity in its
  450. 14:47form makes it a really really good form
  451. 14:50factor. And now I can make one robot
  452. 14:52that does all of these things and I I
  453. 14:54would probably rather have a let's say
  454. 14:56robot that's 5% slower or or you know,
  455. 15:00less reliable than the optimal burger
  456. 15:02flipping robot if I can share
  457. 15:04commonality with that with that nurse.
  458. 15:07In defense, that's not really the case
  459. 15:11because you are in this constant race to
  460. 15:13not just barely beat your adversary, but
  461. 15:15to massively overpower them. To try and
  462. 15:18make it where you know, the goal is is
  463. 15:20to just barely win a conflict on the
  464. 15:22margins. It is to fight unfair, to have
  465. 15:26overwhelming threat of violence
  466. 15:28underpinning whatever diplomatic
  467. 15:30overtures or negotiations you have. And
  468. 15:32and and everyone was always pushing the
  469. 15:34limits. So, if if if I if I have a jet
  470. 15:37that has a reliable 5% every time edge
  471. 15:40over the enemy, that's worth paying a
  472. 15:42lot of money for, right? Like I I I and
  473. 15:45and I I I think the the
  474. 15:48the loss in performance you would see
  475. 15:49through lack of specialization probably
  476. 15:52makes it non-viable. Like to to to go to
  477. 15:54the to the extreme like a you know, I
  478. 15:56build robot submarines and those have a
  479. 15:58specialized form that needs to dive to
  480. 16:006,000 m depth and and sit on the bottom
  481. 16:03of the ocean for months at a time. And
  482. 16:05then I have other stuff that's on the
  483. 16:07surface and and you know, they're
  484. 16:09robotic systems that you know, roll up a
  485. 16:10hill and and they loiter and then I have
  486. 16:12other things that are you know, aerial
  487. 16:13vehicles that are doing very specialized
  488. 16:15tasks and I just can't imagine a
  489. 16:16convergence that would make sense.
  490. 16:18Unless you're telling me that we're
  491. 16:19going to get some cool like like
  492. 16:20humanoids that have like you know,
  493. 16:22on-board fusion reactors and they can
  494. 16:24dive to 6,000 ft depth.
  495. 16:26>> So, Palmer, why aren't why aren't you
  496. 16:27building those? I mean, if anybody, you
  497. 16:30should
  498. 16:31>> [laughter]
  499. 16:32>> Look, you could potentially build such a
  500. 16:33thing, but then very quickly you realize
  501. 16:35that specialization would allow you to
  502. 16:37build something even better. You say,
  503. 16:39"Okay, I've built something that can
  504. 16:40dive to the bottom of the ocean and fly
  505. 16:42to the moon." Man, it could get to the
  506. 16:44moon a lot faster if it didn't have to
  507. 16:46also dive to the bottom of the ocean.
  508. 16:47So, I think I think defense is going to
  509. 16:48be this one where kind of like a
  510. 16:51you know, you you look at for example
  511. 16:52the difference between cars where
  512. 16:54everything has shared platforms within a
  513. 16:56manufacturer you know, manufacturer and
  514. 16:58they also share a lot of the same
  515. 17:00standards
  516. 17:01and the interfaces.
  517. 17:03I think that you compare that with like
  518. 17:05a motorsport where it's all completely
  519. 17:07different and it's continuously pushing
  520. 17:09forward. I I think the defense side is
  521. 17:12probably always going to be this this
  522. 17:14weird branch of exotic highly
  523. 17:16specialized robotic systems.
  524. 17:18>> You know, Palmer, I expected if anybody
  525. 17:20was going to be building humanoid
  526. 17:21robots, it would be you.
  527. 17:22>> No, no, no.
  528. 17:22>> just you you said you said absolutely
  529. 17:24not.
  530. 17:24>> I've I've opted out because um I because
  531. 17:27I believe in them. If I didn't believe
  532. 17:29that humanoid robotics had such a
  533. 17:30powerful future, then I would probably
  534. 17:32build them. But I want Anduril to build
  535. 17:34things that wouldn't exist otherwise.
  536. 17:36That's what I generally want to do.
  537. 17:40Correct. All the other people are
  538. 17:41already doing a great job. Like I I I
  539. 17:44want to use their robots.
  540. 17:45>> And and and do things with them. Like I
  541. 17:47think humanoids are going to be an
  542. 17:48incredible general artificial
  543. 17:51intelligence interface between AI and
  544. 17:54remote systems and a lot of legacy
  545. 17:56platforms that were designed for a
  546. 17:57person to you know pull the lever, push
  547. 18:00the button, read the screen. Like we're
  548. 18:02going to be able to take all those and
  549. 18:03automate them without replacing them
  550. 18:05entirely. So, you have a lot of things
  551. 18:07like
  552. 18:08radar systems that cost millions, tens
  553. 18:10of millions, even hundreds of millions
  554. 18:11of dollars.
  555. 18:12Uh nobody wants to man them during a hot
  556. 18:14conflict cuz they're the first thing
  557. 18:16that's going to go. But if I can put
  558. 18:17humanoid robots in them, I've now been
  559. 18:20able to automate them without that. So,
  560. 18:21yeah, I'm not going to build humanoids.
  561. 18:23Uh this is also why Anduril is not
  562. 18:25building like small quadcopter attack
  563. 18:27drones. Too many other companies doing a
  564. 18:28great job. There's a there's a lot of
  565. 18:30a lot of areas. And and we're not going
  566. 18:31to do space launch. Like we do space
  567. 18:33products. We have a lot we operate a lot
  568. 18:35of our own satellites and a lot of our
  569. 18:36own sensors in space. Uh but I don't
  570. 18:38want to be a launch provider. That's for
  571. 18:39all the other billionaires to
  572. 18:41>> Yeah, yeah, let let them handle that.
  573. 18:42That's fantastic. Salim,
  574. 18:44um you know, when you're operating
  575. 18:47Anduril Industries, you're uh
  576. 18:49decentralizing defense, right? The
  577. 18:52Pentagon and ministries of defense
  578. 18:55globally have been the most centralized
  579. 18:57entity in the history of humanity.
  580. 18:59Uh you're sitting on the edge dealing
  581. 19:01with that. Uh
  582. 19:03you must have a massive immune system
  583. 19:05problem where people inside the entity
  584. 19:07are going, "Nah, we've done that before.
  585. 19:09We can do it better." They either try
  586. 19:11and suck you in or they try and spit you
  587. 19:13out. How are you navigating that
  588. 19:15interface because you're doing it in a
  589. 19:16way that I've never seen anybody succeed
  590. 19:19do it.
  591. 19:22>> I think that this is a problem that's
  592. 19:23very, very hard at the beginning.
  593. 19:26>> Oh, I said the wrong thing.
  594. 19:28>> [laughter]
  595. 19:28>> Wow.
  596. 19:29>> No idea what the
  597. 19:30>> Well, until someone gets on the intercom
  598. 19:31and tells us to
  599. 19:33>> Someone won a million dollars.
  600. 19:34Congratulations.
  601. 19:36>> Um what were we talking about? What what
  602. 19:38what what was it?
  603. 19:44Seriously, I knew the intercom was going
  604. 19:46to come on.
  605. 19:47What were we talking about again?
  606. 19:48>> Centralization of Pentagon versus
  607. 19:50decentralized government.
  608. 19:52>> I'm going to I'm going to keep talking
  609. 19:53till somebody tells me that I can't. Um
  610. 19:55So, we've gotten through this a bit. I
  611. 19:57think early on there was that concern
  612. 19:59that Anduril was not going to be able to
  613. 20:01get through that. Our Our hardest
  614. 20:03problem raising money
  615. 20:04>> Skippy, would you please shut this down?
  616. 20:06>> [laughter]
  617. 20:08>> I've been through enough of these to to
  618. 20:09know how it's going to work.
  619. 20:11>> There we go.
  620. 20:11>> I imagine.
  621. 20:12>> One of our early problems was that our
  622. 20:14investors said, "Palmer, you potential
  623. 20:16investors." They said, "Look, we believe
  624. 20:18that you'll be able to build things that
  625. 20:19work. We believe you'll be able to build
  626. 20:21products that are great, but we don't
  627. 20:23believe the government is capable of
  628. 20:25actually purchasing them at scale." Cuz
  629. 20:26I mean, you talk you talk about
  630. 20:27centralization, you know about the
  631. 20:29issues. For people who don't, uh until
  632. 20:32just a few years ago, you had about 80%
  633. 20:35of MDAPs, major defense acquisition
  634. 20:38programs. So, the basically anything
  635. 20:41that that that matters, uh 80% of MDAPs
  636. 20:44went to five companies, and that's it.
  637. 20:48And I mean, you've seen incredible
  638. 20:50consolidation, centralization. Anduril
  639. 20:52has been successful because we use we're
  640. 20:54we're we're less of a contractor and
  641. 20:55more of a product company. We use our
  642. 20:57own money to build products, then we
  643. 20:58sell them, and we've gotten to the point
  644. 21:00now where people in government have done
  645. 21:02very, very well betting on Anduril, and
  646. 21:04then miraculously we finish on schedule,
  647. 21:07we finish on time, and then the next
  648. 21:09program manager says, "I know how I can
  649. 21:11make my job easy. I just I just give the
  650. 21:14contract to Anduril."
  651. 21:15>> Got it.
  652. 21:16>> So, it it's got to take guts to enter
  653. 21:18the defense industry, right? The most
  654. 21:20incestuous,
  655. 21:21you know, self-licking ice cream cone on
  656. 21:23the planet.
  657. 21:24>> Yeah.
  658. 21:25>> And and and what insanity,
  659. 21:28you know, gave you the balls to do that?
  660. 21:31>> The insanity was the reason that I got
  661. 21:33into it.
  662. 21:35Uh the first page of our pitch deck said
  663. 21:37Anduril will save taxpayers hundreds of
  664. 21:39billions of dollars a year by making
  665. 21:42tens of billions of dollars a year. The
  666. 21:43idea being we're not just making
  667. 21:45products that we're selling, but we are
  668. 21:46displacing the systems that cost
  669. 21:49needlessly 10 times what they should.
  670. 21:52And
  671. 21:53Yeah, thank you.
  672. 21:54>> [applause]
  673. 21:56>> And so, it does seem crazy, but, you
  674. 21:58know,
  675. 21:59I don't I I I can't put myself on the
  676. 22:01same pedestal as him, but one of my
  677. 22:03mentors and idols growing up and then
  678. 22:06also early in my career was John
  679. 22:07Carmack, the founder of id Software.
  680. 22:09>> John is great.
  681. 22:10>> John is incredible. And for people who
  682. 22:11don't know, he's basically the inventor
  683. 22:13of the modern modern 3D game engine,
  684. 22:16first-person shooters, also started a
  685. 22:18rocket company that competed in the
  686. 22:20>> He he was competing in the X Prize, and
  687. 22:21I I remember
  688. 22:22>> And he was one of the, if not the only
  689. 22:25one to be doing reusable
  690. 22:27>> Exactly, he was doing vertical takeoff,
  691. 22:28vertical landing.
  692. 22:29>> else was even trying to do. He did it.
  693. 22:31>> I remember there's this this name on the
  694. 22:33side of his vehicles, Armadillo
  695. 22:35Aerospace.
  696. 22:36How do you pronounce that thing? I
  697. 22:37thought you'd only know. That's right.
  698. 22:38He was using So, yeah, he was using uh
  699. 22:39he was using uh he was building his own
  700. 22:40sensor fusion algorithms. He was
  701. 22:41processing them on modern, you know, or
  702. 22:43on you know, commercial-style graphics
  703. 22:45card hardware using his game programming
  704. 22:50experience. Anyway, the point is it was
  705. 22:52called Armadillo Aerospace. That's
  706. 22:55right. And the thing and yeah, they were
  707. 22:57doing vertical takeoff and landing long
  708. 22:59before SpaceX was
  709. 23:01was was was I mean, doing anything.
  710. 23:03>> Yeah. No, before it existed, for sure.
  711. 23:06>> And but anyway, the the point is he he
  712. 23:10I I he ended up becoming the CTO of
  713. 23:12Oculus VR. I managed to poach him away
  714. 23:14from Id, but after he left, he went to
  715. 23:18work on AI actually even before I did.
  716. 23:22And so he started he started his own
  717. 23:23company to work on on on AGI. And
  718. 23:27and I said why why are you why are you
  719. 23:29doing this?
  720. 23:30>> You know, that the odds of you
  721. 23:30succeeding and to be clear, I still
  722. 23:32believe AI was going somewhere, but he
  723. 23:34said, "I'm going to start a company
  724. 23:35that's just going all in on the end
  725. 23:38state. Like I'm not trying to
  726. 23:39incrementally get there. It's just all
  727. 23:40in on AGI." And the point that he made
  728. 23:42was, "Palmer, even if I've got a 1%
  729. 23:45chance of succeeding, which maybe is an
  730. 23:47overestimation, but suppose I have a 1%
  731. 23:49chance of succeeding, and suppose that I
  732. 23:51could pull it off,
  733. 23:55the the risk-adjusted
  734. 23:57outcome kind of demands that I do it.
  735. 24:01And that was how I felt about defense.
  736. 24:02If there was a 1% chance that I could
  737. 24:05help solve this problem, then it was one
  738. 24:07that would be massively impactful to
  739. 24:09work on versus, let's say, starting
  740. 24:11another gaming company, getting into
  741. 24:13space launch,
  742. 24:15you know, building a car company, which
  743. 24:18it it all of the billionaires, they end
  744. 24:20up doing the same thing. You don't have
  745. 24:21enough people getting into defense or or
  746. 24:23putting on a bat suit. It's just it's
  747. 24:26just cars and rockets and and they just
  748. 24:28go in circles like that for for for
  749. 24:31>> So, you know, when I interviewed you on
  750. 24:32the on the Moonshots podcast last time,
  751. 24:34you mentioned that you're part of a
  752. 24:36billionaire's boys club chat.
  753. 24:38>> Yeah, yeah.
  754. 24:39>> [clears throat]
  755. 24:39>> Oh, yeah.
  756. 24:39>> Are they working on anything worthwhile
  757. 24:41at this point? I mean, I gave I gave
  758. 24:43them for saying like, "God, you got
  759. 24:44all this capital. Solve the world's
  760. 24:46biggest problems."
  761. 24:47>> I think the B boys are doing a lot
  762. 24:48better now than we were a few years ago.
  763. 24:50>> Okay.
  764. 24:51>> Um
  765. 24:51>> They're they're stepping up.
  766. 24:53>> And and for people you think we're
  767. 24:54talking about is I I was giving a lot of
  768. 24:56people a hard time, especially some
  769. 24:57retired B boys. And uh I was telling
  770. 25:01them what are you what are you doing?
  771. 25:01You have cuz money is one of the
  772. 25:04resources they have. But one of the
  773. 25:06resources people don't talk about,
  774. 25:07especially for the retired B boys, is
  775. 25:09that they have credibility, which would
  776. 25:11allow them to gather people around them,
  777. 25:13teams around them, and capital. Like
  778. 25:16they could and it's just sitting there
  779. 25:17underutilized if they're doing nothing.
  780. 25:20They have the credibility in that people
  781. 25:22trust them as a leader to be able to
  782. 25:23lead an organization like that.
  783. 25:26They could have the same ideas and the
  784. 25:28same words as somebody who's equally
  785. 25:30smart or smarter and they'll still do
  786. 25:33better for that. And so my point to them
  787. 25:34is you have a responsibility not only as
  788. 25:36someone who's who's got a lot of capital
  789. 25:38yourself, but who's got this
  790. 25:39underutilized resource of the ability to
  791. 25:41do more.
  792. 25:44If if you're not doing that, then are
  793. 25:46you really like what are you even doing?
  794. 25:48Anyone can sit on a boat. Few people can
  795. 25:51start another multi-billion dollar
  796. 25:53company that does something that
  797. 25:54matters.
  798. 25:54>> [applause]
  799. 25:56>> And the good news is most of them I
  800. 25:58think I I I think a lot of people have
  801. 26:00have gotten the itch to to solve some of
  802. 26:02these problems that are out there.
  803. 26:04Nihilism is is is on is on the down.
  804. 26:10>> [laughter]
  805. 26:10>> Yeah, that's an interesting point.
  806. 26:12Although
  807. 26:13Like I said, I think most of the B boys
  808. 26:15I think they seem like they keep pretty
  809. 26:17busy.
  810. 26:18I don't I don't I don't have a problem
  811. 26:20with B boys who don't do enough. Like to
  812. 26:22some people they mis- mistake what I
  813. 26:23what I say like oh Palmer wants other
  814. 26:25billionaires to work on defense or
  815. 26:27energy or or biology.
  816. 26:30You like
  817. 26:31my point is not that people need to
  818. 26:33distract themselves by doing more
  819. 26:34things. A lot of people can only do one
  820. 26:36thing.
  821. 26:37You know, and that was what I did for a
  822. 26:38long time.
  823. 26:40It's what I hate is the people who are
  824. 26:42not doing anything. That's that that is
  825. 26:44that is like a repulsive to me.
  826. 26:46>> So maybe message to the audience, you
  827. 26:48get to join the B boys slack group or
  828. 26:50whatever this is,
  829. 26:51Telegram, if you're doing something new
  830. 26:54and differentiated.
  831. 26:55>> The group actually got shut down and uh
  832. 26:57we've been we've been talking about now
  833. 26:59that we need to do a B boys 2.0.
  834. 27:01Um but I think I think people post
  835. 27:03>> some B girls as well.
  836. 27:04>> I think people posted the wrong memes a
  837. 27:05few too many times.
  838. 27:07>> Boom.
  839. 27:08>> I I look uh there there was there was a
  840. 27:10time where there were no there were no B
  841. 27:12girls and then there was this Business
  842. 27:14Insider article uh whining about it and
  843. 27:17uh the the reality is like it's just
  844. 27:18like me and some guys.
  845. 27:20>> Like you know, it's not like there's
  846. 27:21this isn't like the Illuminati
  847. 27:22conference.
  848. 27:23>> [laughter]
  849. 27:23>> You know, it's like they're they're like
  850. 27:25oh, there's a group and you know, but no
  851. 27:27women have been invited. I'm like I it's
  852. 27:28not that no women have been invited.
  853. 27:30It's just like me and some some of my
  854. 27:32boys. But
  855. 27:33>> [laughter]
  856. 27:33>> uh but actually there there was a there
  857. 27:35was a there was a there was a girl in it
  858. 27:37by the end. So, uh we need Yeah, we need
  859. 27:39to we need to restart that. We need to
  860. 27:40restart that group.
  861. 27:41>> So, you're in Facebook commercials now.
  862. 27:44>> Oh, yeah. Yeah, yeah.
  863. 27:45>> Yeah, so so the new Meta VR is out, uh
  864. 27:49how does it compare against your vision
  865. 27:50of what you wanted to achieve? Is it
  866. 27:53Yeah.
  867. 27:54>> Look, it does a lot of the things that
  868. 27:56uh it's not exactly what I would make
  869. 27:57were I the guy who's waving the magic
  870. 27:59wand and and and and you got lots of
  871. 28:01magic wands spinning the CAD, but
  872. 28:04uh
  873. 28:04look, the new Meta VR glasses are a lot
  874. 28:07closer in a lot of ways. Uh so, I've
  875. 28:09long been a believer that you need to
  876. 28:11decouple the compute and the battery
  877. 28:14from the head. You should it should not
  878. 28:16be that and and I'm not even saying the
  879. 28:18head totally, it's more of the front.
  880. 28:19So, uh before Oculus Quest. So,
  881. 28:22Oculus Quest was the last thing that I
  882. 28:24worked on before they fired me. And uh
  883. 28:26the prototypes at the time were
  884. 28:28a prototype series code named Santa Cruz
  885. 28:31and uh we were sure actually showing
  886. 28:33that publicly before they fired me. And
  887. 28:36uh Santa Cruz had the battery and
  888. 28:37compute on the back of the head and then
  889. 28:40the light, you know, the part on the
  890. 28:41soft tissue on the front of your head
  891. 28:42was kept as light as possible. And then
  892. 28:44at some point after they got rid of me,
  893. 28:46they they jammed it all into one big
  894. 28:48giant lump in the front and they ruined
  895. 28:49the ergonomics. They forgot why you need
  896. 28:51a top strap. Uh,
  897. 28:53but you know, it it it it bothers
  898. 28:55>> Serve them right.
  899. 28:56>> It bothers me most Look, it bothers me
  900. 28:59mostly at this point because uh
  901. 29:01the Oculus Rift CV1, which was the first
  902. 29:03modern consumer VR headset, it got the
  903. 29:06ergonomics more or less perfect in 2016
  904. 29:10and you know, we we were showing it off
  905. 29:11in 2015 and literally nobody has done as
  906. 29:16good of as it and it's been over a
  907. 29:17decade. Every single person screws it
  908. 29:20up, even Apple. And I I I just I don't I
  909. 29:23can't comprehend it. How how could it be
  910. 29:25that the ergonomic design led by a
  911. 29:2819-year-old kid with no experience is
  912. 29:32better than anything that anyone has
  913. 29:34come up with since. Anyway, the new Meta
  914. 29:36VR glasses are the closest thing.
  915. 29:39And they're very lightweight. They've
  916. 29:40off-boarded the computer. They've
  917. 29:41off-boarded the batteries. They have
  918. 29:43really high-quality pass-through.
  919. 29:45Um they they It has a feature that I
  920. 29:47invented for CV1 called Transformo
  921. 29:49Fabric, which they do not call it.
  922. 29:51The basically using uh fabric uh a a a
  923. 29:55a soft flexible fab black black
  924. 29:58light-absorbing matte fabric to one
  925. 30:00minimize reflections, but two uh
  926. 30:02basically seal the cavity between and
  927. 30:04around the lenses. That means that you
  928. 30:06can move them uh and the fabric just
  929. 30:08stretches to meet where the lenses are
  930. 30:10for different interpupillary distances
  931. 30:12rather than having like sliding
  932. 30:13mechanical interfaces and plastic and
  933. 30:16they brought Transformo Fabric back for
  934. 30:18the first time since my my era. So, I'm
  935. 30:20I'm I'm a huge fan and uh you know, it's
  936. 30:22it's a third of the price of Apple's
  937. 30:24product and uh somewhat higher higher
  938. 30:27pixel density. I I think that's pretty
  939. 30:29extraordinary. Oh, and and and 1/8 the
  940. 30:32weight.
  941. 30:32>> Yeah. Amazing.
  942. 30:33>> Pretty amazing. Pretty good.
  943. 30:36>> Let's close on Let's close on this.
  944. 30:38We've got, you know, 1,500
  945. 30:39entrepreneurs, moonshot visionaries, and
  946. 30:43tens of thousands watching on the live
  947. 30:44stream.
  948. 30:46Uh
  949. 30:46what's your advice?
  950. 30:48What, you know, what did you learn in
  951. 30:50high school that made you succeed at the
  952. 30:54speed you did?
  953. 30:55We were talking about this on the plane
  954. 30:56right back.
  955. 30:57>> I I was really lucky to uh I guess maybe
  956. 31:00not lucky. I did it. Uh but
  957. 31:02>> Well, you are you are lucky.
  958. 31:04>> That's right.
  959. 31:05>> I am [laughter] lucky.
  960. 31:06>> It's in the name.
  961. 31:06>> I started an online community called
  962. 31:09ModRetro when I was 14 or 15. Uh it was
  963. 31:12an internet forum for people who were
  964. 31:15modifying vintage game consoles and
  965. 31:17computers, building handheld units,
  966. 31:19improving the performance in a variety
  967. 31:21of ways. We had a few thousand active
  968. 31:22members.
  969. 31:24And uh and a whole team of volunteer
  970. 31:26moderators and administrators with with
  971. 31:28me me at me at the top of it for some
  972. 31:30reason.
  973. 31:31And uh I learned a lot of valuable
  974. 31:33lessons about how you can build an
  975. 31:35organization, how you market an
  976. 31:37organization, how you get people to
  977. 31:39believe in what you're doing. I learned
  978. 31:41a lot about dealing with the types of
  979. 31:44problems you deal with when you're
  980. 31:45running a company, you know.
  981. 31:46>> With humans.
  982. 31:47>> Well, I mean, you you have you have two
  983. 31:48It's a tale as old as time. You've got
  984. 31:50two people.
  985. 31:51They're both brilliant. They're both
  986. 31:53very good at what they do.
  987. 31:54And they hate each other. And they
  988. 31:56disagree over something. You need to
  989. 31:58somehow mediate the conflict or separate
  990. 32:02them in a way that allows you to get the
  991. 32:03value that that allows both of them to
  992. 32:06contribute value to the greater project,
  993. 32:08uh but not have them hate you, not have
  994. 32:11them lose faith in the broader thing,
  995. 32:12but also you not not kill each other.
  996. 32:15Sometimes that means you need to
  997. 32:16reconcile things between the two.
  998. 32:18Sometimes it means that you need to put
  999. 32:20one guy over here and one guy over here.
  1000. 32:22And sometimes you need to make sure that
  1001. 32:23they don't know that that's what you're
  1002. 32:24doing as you do it. That's just
  1003. 32:26[laughter] I'm not saying that's the key
  1004. 32:27to running a successful business is the
  1005. 32:29management of the two problem,
  1006. 32:31but it's
  1007. 32:33Yeah, and then and once you make it a
  1008. 32:34three-body problem, now I'm out.
  1009. 32:36>> [laughter]
  1010. 32:37>> But uh but but it's it's that and a
  1011. 32:39hundred things like it that I learned
  1012. 32:42running Mod Retro that I was able to
  1013. 32:44take into starting Oculus, to carry
  1014. 32:46forward into what I'm doing in my other
  1015. 32:48companies. And uh that that's why I'm a
  1016. 32:50I'm a strong believer that the the the
  1017. 32:52kids should should should get a job. And
  1018. 32:54I I I also had a job starting when I was
  1019. 32:5614, so I learned how the real world
  1020. 32:57works, too. But I I So, I'm a huge fan
  1021. 33:00of people doing things at a scale that's
  1022. 33:02larger than just themselves. In other
  1023. 33:04words,
  1024. 33:05don't just do a project, do a project
  1025. 33:07with someone else. Do a project with
  1026. 33:09other people. Build in the open, learn
  1027. 33:12the the the the good parts of that, the
  1028. 33:14bad parts of that. It it it's been
  1029. 33:16extremely useful to me. And
  1030. 33:18>> Has building changed in the age of AI?
  1031. 33:19What advice do you have? Again, you
  1032. 33:21know, we've been saying on the pod for a
  1033. 33:23while now, you know, forget about grad
  1034. 33:25school, you know, even college is a is a
  1035. 33:27maybe. You didn't go to college. You
  1036. 33:29know, and high schoolers should what?
  1037. 33:32Find a a purpose they care about and use
  1038. 33:35AI to go and code it. What's your your
  1039. 33:37closing advice here on that?
  1040. 33:40>> You know, one of the reasons I have not
  1041. 33:42been as upset as some people about uh
  1042. 33:48Now, it's funny. There's something on
  1043. 33:49the screen. It's like it's to not talk
  1044. 33:51about
  1045. 33:52>> [laughter]
  1046. 33:53>> Just wanted to make sure you saw that.
  1047. 33:54>> I I I did. We're not going to talk about
  1048. 33:55Don't think about it. All right,
  1049. 33:56company. All right, great. Great.
  1050. 33:59>> [laughter]
  1051. 34:00>> The uh
  1052. 34:02There's the There's a lot of people who
  1053. 34:03who who love their craft and they love
  1054. 34:05what they do. And uh and and they are
  1055. 34:08rightfully mourning uh the end of, for
  1056. 34:11example, coding things by hand top to
  1057. 34:13bottom, start to finish as anything but
  1058. 34:15like a craft endeavor. Um and and I and
  1059. 34:18I understand that. It's it's the same
  1060. 34:19way that a lot of artists felt when the
  1061. 34:21camera was invented. They They whined
  1062. 34:23and complained about how it ruined art
  1063. 34:25and it made it so that anybody could
  1064. 34:26push a button. And and you know, the
  1065. 34:28these things are always painful for the
  1066. 34:29people who are masters of their craft.
  1067. 34:32Um I already got screwed in that way a
  1068. 34:34long time ago. When you're a founder,
  1069. 34:35you don't get to actually work in any of
  1070. 34:36the tech anyway. All the other people
  1071. 34:38get to have all the fun. And so, you
  1072. 34:40know, I I like electrical engineering. I
  1073. 34:42like optical engineering. I like
  1074. 34:43mechanical engineering. I don't get to
  1075. 34:45do any of it cuz the job of a founder is
  1076. 34:47to
  1077. 34:48hire people to to displace them. And and
  1078. 34:51and to do what they thought they could
  1079. 34:53do or what they like they do like doing
  1080. 34:56better. The job, I've said many times to
  1081. 34:58people, if you want to work on
  1082. 34:59technology
  1083. 35:01work for somebody else. If you want to
  1084. 35:03work on nonstop start a
  1085. 35:05company.
  1086. 35:07And so and so what I would say to people
  1087. 35:08is AI doesn't actually change the
  1088. 35:11part of the job very much.
  1089. 35:13You're still on the hook legally. You're
  1090. 35:15still on the hook for all the
  1091. 35:16interpersonal stuff. And you could you
  1092. 35:18you can delegate some of it, but at the
  1093. 35:19end of the day it's your call, it's your
  1094. 35:21decision. It's the same thing for
  1095. 35:23fundraising. It's the same thing for
  1096. 35:24general product architecture. And when
  1097. 35:26people make the wrong decisions, it's
  1098. 35:27your choice
  1099. 35:29as to when you need to overrule them,
  1100. 35:31override them, replace them. I don't see
  1101. 35:33AI replacing that part of it anytime
  1102. 35:36soon. Not because it couldn't, but
  1103. 35:38because the gains everywhere else are so
  1104. 35:40much larger that I think we've got a
  1105. 35:42good 10 years of speed running
  1106. 35:45all of the other parts of the business
  1107. 35:47before they before they come to that.
  1108. 35:48So, look, AI changes a lot, but I I I'm
  1109. 35:51telling you there's a lot of people you
  1110. 35:52are in for a rude awakening if you are
  1111. 35:54like a hobbyist coder or a hobbyist
  1112. 35:56engineer and you think you're going to
  1113. 35:58start a company and you think you're
  1114. 35:59going to be doing all that stuff? No,
  1115. 36:01you're going to learn a hard lesson. It
  1116. 36:04Being a founder was was always like AI
  1117. 36:06took your job. It's just that other
  1118. 36:08people took it.
  1119. 36:09>> Yeah.
  1120. 36:10Ladies and gentlemen, let's give it up
  1121. 36:12for Palmer Luckey.
  1122. 36:19>> [music]

About this transcript

This page contains the full transcript of Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Humanoids | EP #295 by Peter H. Diamandis, generated from the public captions YouTube serves with the video. The transcript has 7,244 words across 1,122 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.