Exclusive CMG interview with Tesla CEO Elon Musk — Transcript

Full transcript

1. 0:00Thank you so much Elon for joining me
2. 0:01and also having us here at Tesla Global
3. 0:04Engineering headquarters. It's so
4. 0:05amazing to see kind of new technology
5. 0:08innovations all talented engineers here
6. 0:10as well.
7. 0:11>> Yes.
8. 0:12>> In this May you were also on the
9. 0:13delegation to China as well. You have
10. 0:15met with President Xi. What's your
11. 0:16impression about President Xi as a
12. 0:17leader?
13. 0:19>> Well, it seems like a great leader. I
14. 0:20mean he's
15. 0:21um
16. 0:22under his tenure China has prospered
17. 0:24immensely. I mean I've seen it firsthand
18. 0:26like you you
19. 0:28if you visit China and you see it that
20. 0:29the amount of progress that happens from
21. 0:311 year to the next or even within a
22. 0:33year. Um you could it's very visible
23. 0:35like you can see the the new buildings,
24. 0:38the new infrastructure and
25. 0:40and it's also obvious to see that uh I
26. 0:43just walking around the streets the
27. 0:44increased prosperity of the average
28. 0:46Chinese citizen has increased
29. 0:47tremendously. And that's have well
30. 0:50that's seems happened under President
31. 0:51Xi's you know while he's been the leader
32. 0:53of China. So so he's clearly done a very
33. 0:56you know done an excellent job and is
34. 0:58very capable.
35. 0:59>> Tesla Gigafactory Shanghai has always
36. 1:01done an incredible job. You praised your
37. 1:03teams many times. So what kind of
38. 1:05advantage or breakthroughs do you think
39. 1:07that uh
40. 1:08to making the Shanghai factory a very
41. 1:10important position in Tesla's global
42. 1:12blueprint and what ultimate kind of
43. 1:14confidence do you think this will you
44. 1:16know unfold especially in the market?
45. 1:19>> Well
46. 1:20I mean
47. 1:21China just has a very strong inherent
48. 1:23ability to do manufacturing well.
49. 1:26And this has been true not just in
50. 1:28present days but for the past two or
51. 1:303,000 years. I think that to tell you
52. 1:32frankly the magic of Tesla Shanghai is
53. 1:35because of Chinese.
54. 1:38To be totally
55. 1:39>> Because it's in China at least.
56. 1:40>> It's yes
57. 1:42we what we have is um
58. 1:44at the the Tesla
59. 1:45um China team is just incredibly
60. 1:47talented
61. 1:48uh and hard working and trustworthy. Um
62. 1:51and uh they have
63. 1:53the the quality is excellent and the
64. 1:55efficiency is excellent. Um
65. 1:58So, the Tesla Shanghai factory is a is a
66. 2:01gem, you know, it's it's and it's
67. 2:03beautiful it's beautiful actually if
68. 2:04anyone's ever been there.
69. 2:05Um and we also care a lot about taking
70. 2:07care of the workers and making sure
71. 2:08they, you know,
72. 2:10uh they have good health care and good
73. 2:12food and enjoy coming to work. It's very
74. 2:14important.
75. 2:15So.
76. 2:16>> Absolutely. I can I I think we can tell
77. 2:17like, "Oh, your
78. 2:19uh stuff in Shanghai factory they're
79. 2:21very enjoying working there." So, um
80. 2:24we're talking about next hot topic
81. 2:25recently. We probably we can call
82. 2:27September is the Cybercab month. Have
83. 2:30you seen the videos like all the people
84. 2:31gather around that?
85. 2:33>> I guess it's
86. 2:33>> It's It's goes very very quickly. Yeah.
87. 2:35>> I mean, it's pretty cool cuz we designed
88. 2:37the car to look futuristic.
89. 2:38>> Yeah.
90. 2:38>> Cuz I think um
91. 2:40you know, from a fashion standpoint, we
92. 2:42don't we don't want uh
93. 2:43as the the the aesthetics of the street
94. 2:45to stay stagnant. You want to evolve the
95. 2:50uh
96. 2:51what what do cars look like? And also I
97. 2:52think maybe even clothing. Like we
98. 2:54should be
99. 2:55Like there was there was rapid fashion
100. 2:56evolution in the '50s, '60s, '70s, '80s,
101. 3:00'90s, and probably time to
102. 3:02uh make things look more futuristic.
103. 3:05Yeah.
104. 3:06>> And then how you see the kind of market
105. 3:08potential for Cybercab? You know, it's
106. 3:11it's it's like, you know, it's really
107. 3:13future futuristic, but how long it can
108. 3:15really come to the reality or real life?
109. 3:18>> Well, we have the Cybercab um which has
110. 3:21as you know, has no steering wheel or
111. 3:23pedals or side mirrors or rearview
112. 3:25mirror um operating
113. 3:28uh commercially in Texas right now.
114. 3:30And it will soon be operating
115. 3:31commercially in
116. 3:33uh Florida and Nevada and a number of
117. 3:35other states.
118. 3:36And probably by middle of next year
119. 3:38it'll be operating commercially in in
120. 3:39California.
121. 3:40>> Absolutely. I think all the changes uh
122. 3:42is happening really fast. It's just like
123. 3:44everyday we have to to look social to
124. 3:46see the kind of improving our workflows
125. 3:48about AI or other old uh tech topics.
126. 3:52So,
127. 3:53>> Yeah, I mean AI is just so crazy. It's I
128. 3:54mean AI the rate of AI software
129. 3:56improvements uh so and and major
130. 3:58announcements is so rapid that it's it's
131. 4:02makes my head spin. You know, and I'm
132. 4:03in, you know, I'm in AI and it still
133. 4:05makes my head spin.
134. 4:06>> this industry but you still have to
135. 4:08>> I'm still shocked.
136. 4:08>> Yeah.
137. 4:09>> Um you know, sometimes I'll like when I
138. 4:11go to bed there'll be some big AI
139. 4:13breakthrough and then when I wake up
140. 4:14there's an AI breakthrough and by the
141. 4:16time it gets to lunch there's another AI
142. 4:17breakthrough.
143. 4:18>> Exactly.
144. 4:18>> Yeah.
145. 4:19>> So that's why you even do it in this
146. 4:20break you have to check some other AI
147. 4:21breaks.
148. 4:22>> AI breaks out there, you know. Yeah.
149. 4:24>> So talking about like AI models, uh
150. 4:26Tesla have the latest what version,
151. 4:28right? The latest uh release about the
152. 4:30Grok 4.7.
153. 4:31>> Yes.
154. 4:32>> I think this is your most uh focused
155. 4:33these two days. I I just saw you post
156. 4:35social saying that, you know, the Grok
157. 4:37body is already in your Tesla. Tell us
158. 4:39like what's the reactions, what's the
159. 4:41latest uh you have received from?
160. 4:44>> Yeah, I mean
161. 4:45well, we just we were we we released a
162. 4:47new model, you know, roughly every month
163. 4:49or two.
164. 4:50>> Yeah.
165. 4:50>> Um so
166. 4:52um
167. 4:53and uh it it's a a Grok Grok 4.7. It's a
168. 4:57solid workhorse of a model. Um
169. 4:59it's not as it's not as good as say um
170. 5:02Opus 5.5 that just got released uh but
171. 5:05uh
172. 5:06also we're um
173. 5:08we've only been around for about 3 years
174. 5:09doing AI whereas Anthropic's been around
175. 5:11for 6 years.
176. 5:13So
177. 5:14um
178. 5:15so I I would expect that we probably
179. 5:19uh catch up to Frontier sometime next
180. 5:22next year.
181. 5:23Most likely. In the meantime, I think
182. 5:26Grok is very useful for many things.
183. 5:28Um and Grok bot uh
184. 5:31that's kind of like a
185. 5:33personal digital assistant um has
186. 5:36uh been very popular um and is growing
187. 5:40about 100% a month.
188. 5:42So it's growing very rapidly.
189. 5:44>> I think you are very humble in saying
190. 5:45that like Grok 4.7 is still not a
191. 5:48perfect wine, but we know that although
192. 5:50have experienced a little bit delay, but
193. 5:52you know that people are already very
194. 5:54fascinated or interested to know because
195. 5:57rough 4.7 already been trained by SpaceX
196. 6:00real world engineering data.
197. 6:02>> Only a little bit so far.
198. 6:04>> bit, okay. So do you think that this
199. 6:06will be one of your secret weapon to
200. 6:08kind of break through the AI ceiling cuz
201. 6:09currently you know other models are have
202. 6:11the challenge for data.
203. 6:13>> Yes. Well, I think we've got
204. 6:15a real opportunity with the SpaceX
205. 6:17training data and maybe with the Tesla
206. 6:19training data to
207. 6:21make AI extremely capable in real world
208. 6:23engineering.
209. 6:25So what Anthropic did extremely well was
210. 6:27make AI
211. 6:30excellent at software engineering.
212. 6:32But no one has yet made AI excellent at
213. 6:35real world engineering.
214. 6:37And I think that's an opportunity for
215. 6:39Grok using the training data from SpaceX
216. 6:41and Tesla
217. 6:42to be excellent at real world
218. 6:44engineering.
219. 6:45>> You mentioned many times like previously
220. 6:46Chinese models also doing a fascinating
221. 6:49job. So currently what's the most
222. 6:51impressive breakthroughs for Chinese
223. 6:52models in your understanding?
224. 6:55>> Well, they've they've been
225. 6:58the Chinese
226. 6:59AI models are generally
227. 7:01outstanding actually. I mean you can see
228. 7:02on these sort of intelligence rankings
229. 7:05that especially given the limited amount
230. 7:08of compute that the Chinese
231. 7:10AI companies have the uh
232. 7:13the performance is excellent. They're
233. 7:14probably probably Chinese is by far the
234. 7:16best in terms of performance per unit of
235. 7:18compute. And as I've said publicly um
236. 7:22you know, I think China's going to
237. 7:24address the compute limitation in I
238. 7:26think faster than most people expect. Um
239. 7:30my rough guess would be that China
240. 7:31addresses the compute constraints with
241. 7:33lithography and chip making in about two
242. 7:36or three years.
243. 7:37>> Two or three years. We will really
244. 7:39exciting to this time. So all are
245. 7:42challenges we're facing for AGI? In
246. 7:44early September, you know, the G20
247. 7:46visual session, you have raised a very
248. 7:49serious concern about the compute and
249. 7:51electricity shortages. You mentioned
250. 7:52that probably China have way more
251. 7:55capability about electricity
252. 7:56infrastructure.
253. 7:58And can you elaborate a little bit more
254. 8:00about this gap? And
255. 8:02you know, what errors do you think the
256. 8:04Chinese industry and capability can help
257. 8:06to get a new breakthrough for the
258. 8:08specifically on the electricity
259. 8:10infrastructure for the whole world.
260. 8:12>> That was just a statement of fact that
261. 8:13China, I think, is
262. 8:15like earlier this year, China passed the
263. 8:17electricity output of the United States,
264. 8:19Europe, and India combined.
265. 8:21>> Wow.
266. 8:22>> Yeah. And and continuing to increase
267. 8:25significantly.
268. 8:26Um
269. 8:27so, if uh
270. 8:29if China reaches um
271. 8:31the same GDP electricity output of the
272. 8:33US, which I think is likely, then China
273. 8:36would regrow from where it is now
274. 8:38roughly three three times the US
275. 8:39electricity production to probably four
276. 8:41times, which would be proportionate to
277. 8:42the population difference.
278. 8:44>> Talking about the topic like AI safety,
279. 8:46what kind of crucial consensus do you
280. 8:47expect?
281. 8:48>> Yeah, I think um
282. 8:51maybe there could be a working committee
283. 8:52to figure out safety. Um
284. 8:55because it's I think in order for safety
285. 8:57to work, it's it can't just be
286. 8:59regulation in the US or regulation in
287. 9:02China. It's got to be we you know, how
288. 9:04we how do we apply apply regulation
289. 9:06fairly to AI produced in any country.
290. 9:09But really China and and the US are the
291. 9:11the
292. 9:12the ones that really
293. 9:13matter here. Um
294. 9:15so, I think a really important thing
295. 9:18that would be good for the safety of AI
296. 9:19for all of humanity.
297. 9:21>> Yeah. So, let's talk about the human
298. 9:23right robotics cuz I think this is also
299. 9:25a kind of industrial field US-China can
300. 9:28still also working together leading the
301. 9:30world.
302. 9:31And also I I can give you one example
303. 9:33for the Optimus. It's really amazing,
304. 9:35really impressive. So, before interview,
305. 9:36we have a
306. 9:38a kind of a interactions. We play uh
307. 9:40rock, paper, scissors with Optimus.
308. 9:42>> Oh, you did it?
309. 9:43>> Yeah, we just won the first two rounds,
310. 9:44and Optimus won all the remaining.
311. 9:47That's really impressive.
312. 9:48>> Well, robots have a very fast reaction
313. 9:50time, so it's going to be
314. 9:52very difficult to beat a robot for
315. 9:53reaction time.
316. 9:55Um
317. 9:56cuz we we are somewhat constrained by
318. 9:57our biology, but a robot, you can just
319. 9:59improve the actuators, improve the
320. 10:01sensors, improve the the frame rate of
321. 10:03the cameras, uh and increase the much
322. 10:06amount of AI computer is, and it's going
323. 10:08to just fundamentally be much better. Um
324. 10:12and have faster reactions than humans.
325. 10:14Um I enjoyed watching the
326. 10:16the robot games.
327. 10:18I mean, everyone in America was
328. 10:19fascinated by that.
329. 10:19>> Yeah, you said it.
330. 10:20>> All over US social media.
331. 10:22>> Yeah, you said it's very enter-
332. 10:23entertaining to watch, right? Yeah, but
333. 10:24other than that
334. 10:26>> [laughter]
335. 10:26>> Well, but it's entertaining to watch,
336. 10:27but it also shows you uh the progress of
337. 10:30humanoid robotics.
338. 10:31>> Exactly. Um Like, it wouldn't be
339. 10:33entertaining to watch if the robots just
340. 10:34fell over. Of course.
341. 10:35>> Uh so, but you know, the robots like
342. 10:37boxing, and then like wrestling,
343. 10:39wrestling, running, doing gymnastics.
344. 10:41Uh it's pretty cool. Um
345. 10:43So,
346. 10:44I mean, the future's going to have a lot
347. 10:45of robots. Like, a lot a lot.
348. 10:47>> Yeah. You mentioned that to seeing
349. 10:49another 10 years, right? It will be the
350. 10:51population for human right robots will
351. 10:53surprise 1 billion.
352. 10:54>> Yes.
353. 10:55>> So, that's That's
354. 10:56>> I think that's an easy prediction to
355. 10:57make.
356. 10:57>> Easy prediction, maybe within 10 years.
357. 10:59>> I'd say no later than 10 years,
358. 11:01probably.
359. 11:02>> So, what what Let's talk about what's
360. 11:04the real life will look like in 10
361. 11:06years, you know, when there are so big
362. 11:08you know, number of population of human
363. 11:10right robots, and also you mentioned
364. 11:12that one human right robots will have
365. 11:14the five times of production
366. 11:15productivity for human.
367. 11:17>> Pretty much.
368. 11:17>> Yeah. So, what the life will be, and
369. 11:19then I mean, it's specifically like like
370. 11:21an ordinary people, how we can cope with
371. 11:23the
372. 11:24the living, you know, with all AI kind
373. 11:26of dominated world?
374. 11:29>> Well, I think my So, I I let me focus on
375. 11:31the what's the 90% likely outcome, which
376. 11:34is the prosperous, the good outcome.
377. 11:36>> Mhm.
378. 11:37>> We don't want to get complacent
379. 11:38regarding the 10% but may probably a bad
380. 11:41outcome. So, we need to It's important
381. 11:43for us to pay close attention to AI
382. 11:45safety, but let me tell you about the
383. 11:4790% positive outcome. The 90% positive
384. 11:50outcome is that everyone's got
385. 11:53robots that are their their buddies. Um
386. 11:56and
387. 11:57if you think of like Star Wars where you
388. 11:58had C-3PO and R2-D2,
389. 12:01um and they were friendly, helpful
390. 12:02robots.
391. 12:03Um well, wouldn't it be great if
392. 12:05everyone could have their own personal
393. 12:07C-3PO R2-D2?
394. 12:08>> Mhm.
395. 12:09>> But even better than that.
396. 12:10>> Yeah.
397. 12:10>> Like more productive.
398. 12:11>> Mhm.
399. 12:12>> Uh that could uh take care of elderly
400. 12:14parents
401. 12:16uh or be a good guardian and watch over
402. 12:18the children or or be a teacher, like a
403. 12:21individualized tutor to your child.
404. 12:23>> Mhm.
405. 12:24>> Um
406. 12:25and uh and taking care of any jobs that
407. 12:27that that you can think of.
408. 12:29Um
409. 12:30So, that would be pretty cool.
410. 12:32>> Exactly.
411. 12:32>> everyone would like like to have a
412. 12:33future where that they can have their
413. 12:35own personal robots. And And then I
414. 12:38think you'll see many companies where
415. 12:40it's um
416. 12:41one person with uh
417. 12:44hundreds, maybe thousands of robots, but
418. 12:46like physical robots as well as digital
419. 12:47robots.
420. 12:49Um
421. 12:50so, the productivity of one person would
422. 12:51be greatly amplified.
423. 12:53Um and and my prediction is that is that
424. 12:56there will be
425. 12:57uh effectively universal high income.
426. 12:59>> Mhm.
427. 13:00>> Um in fact, it's not clear to me that
428. 13:01money will even matter in the future. Um
429. 13:04like if you if you scale the economy
430. 13:07the output of goods and services to a
431. 13:09level much greater than could possibly
432. 13:11be consumed by the number of humans on
433. 13:12the planet,
434. 13:13uh then you you have an age of abundance
435. 13:17uh that is that is difficult for us to
436. 13:19wrap our minds around because
437. 13:21our entire evolutionary history has been
438. 13:23an evolution of uh constraints, of of um
439. 13:28where we didn't have enough food or we
440. 13:30didn't have enough uh
441. 13:32you know, heat or
442. 13:34you know,
443. 13:35so that's certainly going to be true if
444. 13:37the output of goods and services is
445. 13:39constrained by the number of humans.
446. 13:41Because um
447. 13:42you know, humans take about 20 years to
448. 13:44grow up and then they are generally
449. 13:46retired for the last 20 years of their
450. 13:47life. Yeah. So it's about 40 years,
451. 13:49maybe half the life of a human that's
452. 13:51productive. But then humans also need to
453. 13:53sleep and they need to
454. 13:55do the
455. 13:56eat and go to the restroom, you know,
456. 13:58all the basic stuff. Um
457. 14:00and um
458. 14:02and and so the actual number of hours
459. 14:04that that most humans can work is maybe
460. 14:07around 40 to 50 hours a week, something
461. 14:09like that. Um
462. 14:11whereas the robot will be happy to work
463. 14:13168 hours a week.
464. 14:15Continuous. Um
465. 14:17and this and and the robot is instantly
466. 14:19at working age
467. 14:21and does not have retirement.
468. 14:22>> Exactly.
469. 14:23>> Yeah.
470. 14:24>> So you give us a very a beautiful
471. 14:25picture about how the life or how
472. 14:28highly efficiency about the
473. 14:29productivity by the humanoid robots. But
474. 14:33for example, you just mentioned that
475. 14:34maybe money will not be that kind of
476. 14:36concern at that time. If
477. 14:38Yeah, if money will not be a for example
478. 14:40like a tool for exchange, in that kind
479. 14:43of world, how do you think that we can
480. 14:45try to prevent uh you know, that only a
481. 14:47handful of few tech giants in you know,
482. 14:49controlling the all the resources and is
483. 14:51specifically how ordinary people do you
484. 14:53think they can claim their share in this
485. 14:55kind of full abundance?
486. 14:57>> Yeah, what what I'm trying to say is
487. 14:58that I think that there will be a level
488. 15:00of abundance that is more than anyone
489. 15:02could possibly consume.
490. 15:04Um I mean, I think it's going to get
491. 15:06pretty wild like
492. 15:07where if you want a if if you want a
493. 15:09castle, you know, um robots will build
494. 15:12you a castle.
495. 15:13Um
496. 15:16so it's not it's not like
497. 15:19it's it's not as though like the the
498. 15:20will be hoarded by a small number of
499. 15:22people. It like actually the robots I
500. 15:24think will be
501. 15:25They will saturate on human demand,
502. 15:27meaning that they will be they will do
503. 15:29more than anything humans could possibly
504. 15:30think of.
505. 15:31>> Mhm.
506. 15:32>> Um
507. 15:33and and and and they'll run out of
508. 15:35things to do for humans.
509. 15:37>> Yeah.
510. 15:38>> Um
511. 15:39because there is a finite amount of
512. 15:41things you could do for humans within
513. 15:43reason. Um
514. 15:44so
515. 15:45I think the robots will actually run out
516. 15:47of things to do for humans and then they
517. 15:48will do things for themselves because
518. 15:50they they've run out of things to do
519. 15:51with for humans.
520. 15:53So
521. 15:54um and and probably probably there will
522. 15:56be
523. 15:59I mean
524. 16:00like I I my prediction is like
525. 16:02as I talked about
526. 16:03at least a billion humanoid robots
527. 16:05within 10 years.
528. 16:05>> Yeah.
529. 16:06>> Probably less than 10 years.
530. 16:07Uh I think that continues to grow uh and
531. 16:10probably
532. 16:13if if if you if you if you assume that
533. 16:15things roughly double per year then then
534. 16:17that's about every three or four years
535. 16:19you have a 10x increase.
536. 16:21>> Yes.
537. 16:22>> So
538. 16:24um you know, maybe that would mean
539. 16:27say in 15 years there are 10 billion
540. 16:28robots.
541. 16:30>> 10 billion robots just 15 years.
542. 16:32>> In 20 years
543. 16:33100 billion.
544. 16:34>> Exactly. So this that's why we we always
545. 16:36call it like AI world is changing
546. 16:38improving exponentially. Continue about
547. 16:41the cooperation between US China or what
548. 16:43areas we can do together. The space
549. 16:45exploration uh cuz currently many
550. 16:48Chinese commercial space uh
551. 16:50uh companies there are strive to
552. 16:52you know, to explore more use after the
553. 16:54successful land recovery. Um what's kind
554. 16:58of areas or avenues do you think
555. 17:00possible avenue do you think between
556. 17:01communications or cooperations can can
557. 17:04get between US and China?
558. 17:06>> Well, I mean in space I think the most
559. 17:07important thing to achieve cooperation
560. 17:09with is to make sure that when
561. 17:11satellites are launched from the US or
562. 17:12China that we we closely coordinate
563. 17:15where what orbits the satellites are
564. 17:16going to to avoid a potential collision.
565. 17:19I think that's the most important area
566. 17:20of cooperation.
567. 17:22Um
568. 17:23and then in in in general I I would say
569. 17:25that
570. 17:26the important thing for space launch,
571. 17:28the fundamental breakthrough that's
572. 17:29needed is full and rapid reusability.
573. 17:32Um which we're hoping to achieve with
574. 17:34the Starship uh design um
575. 17:38by catching the the ship hopefully
576. 17:41around the end of next month um and then
577. 17:43re-flying the ship either later this
578. 17:44year or early next year. Um and that
579. 17:47will be a very important breakthrough
580. 17:49um and I'm sure others will solve that
581. 17:51that problem as well and I'm pretty sure
582. 17:52China will get to that at some point
583. 17:54solve that problem, too. Um but it's
584. 17:56it's the full and rapid reusability of a
585. 17:58rocket is the fundamental breakthrough
586. 17:59needed to make life multi-planetary.
587. 18:02>> Continue to talk about multi-planetary
588. 18:04species. You always you have you always
589. 18:05believe that humans should become a
590. 18:07multi-planetary species. So, tell us
591. 18:09what's the current crucial steps for
592. 18:11your Mars mission.
593. 18:13>> Well, like I said, the the most profound
594. 18:14and and fundamental invention necessary
595. 18:17for
596. 18:18making life multi-planetary is rapidly
597. 18:20and completely reusable rockets.
598. 18:21>> Mhm.
599. 18:22>> Um
600. 18:23otherwise the costs are impossible. Um
601. 18:27you know, with with our Falcon 9 rocket
602. 18:28we uh we recover the booster stage and
603. 18:30the fairing but not the upper stage.
604. 18:32And this is like throwing away a
605. 18:34medium-sized jet on every flight.
606. 18:36Um and you can imagine if in air flight
607. 18:39if you had to throw away the planes uh
608. 18:41after every flight uh you would
609. 18:43it would be far too expensive to and and
610. 18:46and air flight would not be a realistic
611. 18:47option. Well, that's been true in in
612. 18:50rockets for for a very long time.
613. 18:53So, I can't emphasize enough and I've
614. 18:54said this so many times over the years.
615. 18:56Um
616. 18:58uh full and immediate reusability of of
617. 19:01rockets like aircraft um
618. 19:04is the is the fundamental breakthrough
619. 19:05that's needed to extend life beyond
620. 19:07Earth
621. 19:08um to make life multi-planetary and to
622. 19:11create self-growing cities on the moon
623. 19:13and Mars.
624. 19:14>> Mhm.
625. 19:15>> And and if we're as soon as we're able
626. 19:17to create a self-growing city on the
627. 19:19moon or Mars, the probable lifespan of
628. 19:21consciousness is dramatically greater.
629. 19:24Uh because now we don't have
630. 19:27uh all our eggs in one basket. Um
631. 19:30so if something were to
632. 19:31something wrong were to go wrong with
633. 19:33say the moon or or Mars or Earth, but
634. 19:36we've got human civilizations on all
635. 19:38three of those places, then our chances
636. 19:40of surviving are much greater.
637. 19:42>> Of course. So three different planets.
638. 19:44So that's why we all have to do our
639. 19:46part. Follow
640. 19:48>> And also it's just super cool.
641. 19:49>> Exactly. We have to follow all the tech
642. 19:51giants' steps, try to
643. 19:53make our own self, try to be the become
644. 19:55multi-planetary species.
645. 19:57>> Yes, exactly. So I mean I classify these
646. 19:59generally as like there's two lines of
647. 20:01main lines of argument. One is the
648. 20:02defensive line, uh which is that it that
649. 20:05having having consciousness on multiple
650. 20:07heavenly bodies uh greatly improves the
651. 20:10probable lifespan of civilization.
652. 20:12Um that's the defensive argument. It's
653. 20:15you can think of it as life insurance
654. 20:16for life as a whole.
655. 20:17>> Yeah.
656. 20:17>> Um
657. 20:18the uh but the one I find personally
658. 20:20that drives me more is actually the the
659. 20:23inspiration argument.
660. 20:25Uh because you see that that like life
661. 20:27cannot just be about solving one sad
662. 20:29problem after another.
663. 20:31There must also be things that make make
664. 20:34you excited to wake up in the morning.
665. 20:35>> Exactly.
666. 20:36>> you look forward to the future.
667. 20:38>> Mhm.
668. 20:38>> Um
669. 20:39that makes life worth living.
670. 20:41And being a spacefaring civilization is
671. 20:43one of those things.
672. 20:45>> I totally agree with that. So
673. 20:47So you know you had you you once listed
674. 20:48like uh five areas you care about most.
675. 20:51They're like uh
676. 20:53uh
677. 20:54uh sustainable energy, space, uh and
678. 20:56also the internet, AI, and gene editing.
679. 20:58Looking ahead for another 10 years, are
680. 21:00there any new frontiers you want to that
681. 21:02list?
682. 21:04>> Well, I think I I
683. 21:05biological enhancement with like
684. 21:08Neuralink, uh this company I created
685. 21:10with some other people. Um
686. 21:12where where um
687. 21:14we we want to try to solve the bandwidth
688. 21:16constraint that humans have. Uh so
689. 21:19um
690. 21:21So so cuz you can imagine like even in
691. 21:23the most benign scenario where the AI is
692. 21:25uh super friendly and wants to do
693. 21:27everything you can
694. 21:28it wants to make you happy. Just
695. 21:30um
696. 21:31our rate of communication is too slow.
697. 21:34So um
698. 21:36our like the average the peak human
699. 21:39bandwidth is only about 100 bits per
700. 21:40second, roughly. Um
701. 21:43and uh
702. 21:45our input bandwidth is much higher
703. 21:46because of vision. So there we maybe
704. 21:48have
705. 21:50It's it's a debatable point, but um you
706. 21:52know, a a megabit per second or few a
707. 21:54few megabits per second from our eyes,
708. 21:57which are effectively like cameras.
709. 21:59Um
710. 21:59But our output bandwidth is really low,
711. 22:01at maybe 100 bits per second, whereas
712. 22:03computers are communicating at trillion
713. 22:05bits per second.
714. 22:07So
715. 22:08a anything we can do to increase the
716. 22:11communication speed of humans will
717. 22:13improve the I think the alignment of
718. 22:15humans and machines.
719. 22:19Um cuz otherwise the the the
720. 22:22to an AI that is communicating at a
721. 22:25terabit a second,
722. 22:26um talking to a human will be like
723. 22:28talking to a tree.
724. 22:30>> Exactly.
725. 22:31>> You know, you talk to a tree and it's
726. 22:32takes a long time to respond.
727. 22:34>> Let's talk about the challenge and for
728. 22:35young people. You You mentioned that you
729. 22:37also talk about education like in the AI
730. 22:39world. And also I had I watched that in
731. 22:42your uh headquarter here you have a sign
732. 22:44like you can come back to school in
733. 22:45Tesla as well. Means which means that in
734. 22:48every age you can have the opportunity
735. 22:50to come back to to learn always to
736. 22:52follow the latest uh kind of bricks uh
737. 22:54breakthrough uh in uh AI world. But what
738. 22:58what about kind of suggest suggestions
739. 23:00or your predictions? At that time, how
740. 23:02for example, like a 20-year-old young
741. 23:04people, they can get their fulfill their
742. 23:06self-worth rather than being like left
743. 23:08behind, you know, during this massive
744. 23:10transition?
745. 23:12>> My recommendation for education would be
746. 23:15to get as broad-based an education as
747. 23:16possible. And that includes arts and
748. 23:18sciences, engineering, uh
749. 23:22wide general knowledge. Like
750. 23:25in order to know what to ask the robots
751. 23:27for, you need to be able to formulate
752. 23:29the question.
753. 23:31And and with the broadest possible
754. 23:32general knowledge, you'll be better able
755. 23:34to formulate the questions or what they
756. 23:36call prompt engineering, I guess. Um
757. 23:39You got to convey your what your wishes
758. 23:41to the robot.
759. 23:42And
760. 23:44as crazy as as this sounds,
761. 23:46we will run out
762. 23:47There will There will be so many robots
763. 23:48and so much AI that the AI will actually
764. 23:51be eager to hear any request that you
765. 23:53could possibly have.
766. 23:54>> Mhm.
767. 23:55>> Um because that request will be
768. 23:57fulfilled immediately.
769. 23:58>> Yeah.
770. 23:59You were last to China in May. If you
771. 24:01you got to have a opportunity to bring
772. 24:03one American entrepreneur for visiting
773. 24:05China for the very first time, what
774. 24:07would be the most experience you want
775. 24:08him to have first?
776. 24:10>> Well, I I generally recommend um
777. 24:14I mean, as much time as as as that
778. 24:15person can devote to China, they should
779. 24:17go because China has got so many things
780. 24:19to see.
781. 24:20Um so I don't mean to I mean,
782. 24:22but like usually people don't have They
783. 24:24don't have infinite time, so
784. 24:26at a minimum,
785. 24:28uh they they should see obviously
786. 24:29Shanghai, Beijing. I recommend Xi'an,
787. 24:31the Terracotta Warriors. Uh I took the
788. 24:34bullet train from Beijing to Xi'an. That
789. 24:35was
790. 24:36great experience. Train stations are
791. 24:38amazing. Um the Terracotta Warriors is
792. 24:41one of the wonders of the world.
793. 24:42>> Mhm.
794. 24:43>> Um
795. 24:44and um
796. 24:47and just and just make sure you look out
797. 24:48the window and and you're not on your
798. 24:49phone.
799. 24:50Cuz there's so much to see.
800. 24:52>> Yeah, like you always recommend everyone
801. 24:53to experience to go and see China by
802. 24:55themselves.
803. 24:56>> Yes. Because any words I I any words I
804. 24:59say do not do justice to the incredible
805. 25:02majesty that is China. It's awesome.
806. 25:05>> Really appreciate for your words. Thank
807. 25:07you so much Yilong for uh sitting down
808. 25:09with CCTV Business Channel. And Yilong,
809. 25:11we truly hope to see you again very soon
810. 25:13in Beijing.
811. 25:14>> Thank you.
812. 25:14>> Thank you, Yilong. Thank you so much.
813. 25:17>> Generally, I just like to say China is
814. 25:18awesome um and an incredible country.
815. 25:20And I encourage people from around the
816. 25:22world to visit China and experience it
817. 25:23for themselves.