YouTube2Text

Elon Musk on AI: humans will no longer be in control in ten years | The Economist — Transcript

by The Economist · 1,788 words · 333 segments · language en · Watch on YouTube

Full transcript

  1. 0:00I think
  2. 0:01AI may exceed the sum of human
  3. 0:03intelligence
  4. 0:05in about in around 5 years.
  5. 0:07>> In 5 years?
  6. 0:07>> Roughly 5 years is my guess.
  7. 0:10Um
  8. 0:12So
  9. 0:13there really won't be anything that AI
  10. 0:15can
  11. 0:16can't do better than humans apart from
  12. 0:19being human perhaps.
  13. 0:20>> In a more prosaic level
  14. 0:22what will life be like?
  15. 0:25>> The most likely outcome
  16. 0:28is an age of amazing abundance
  17. 0:31uh where
  18. 0:34anyone can have anything they can think
  19. 0:35of.
  20. 0:37This is this is this may sound
  21. 0:38preposterous but
  22. 0:40well here we are in 2026.
  23. 0:43Let's see where we stand in 2036. Now of
  24. 0:45course there are things that could uh
  25. 0:47derail this like a a massive global
  26. 0:50thermonuclear war or something like
  27. 0:51that. Um
  28. 0:53but assuming that there's not World War
  29. 0:55III
  30. 0:56uh I think
  31. 0:58I think we're headed for an age of
  32. 0:59amazing abundance. So this is I guess a
  33. 1:01message of
  34. 1:03of optimism
  35. 1:05and uh excitement about the future.
  36. 1:07>> You're sure that humans will no longer
  37. 1:09be in control in 10 years time?
  38. 1:11>> I think it is unlikely um in the same
  39. 1:14way that
  40. 1:15if you had a vastly
  41. 1:18more like a a
  42. 1:20you know
  43. 1:22the
  44. 1:24like like if if if the difference in the
  45. 1:26intelligence between AI and humans is
  46. 1:30vastly greater than the difference in
  47. 1:31intelligence between AI and chimpanzees
  48. 1:34it's hard to imagine that the
  49. 1:36chimpanzees would be in charge.
  50. 1:38>> I'm really stuck by this.
  51. 1:41>> um
  52. 1:42an evolved form of chimpanzee. So it
  53. 1:45wasn't that long ago that we were
  54. 1:47swinging through the trees eating
  55. 1:48bananas um in the jungle. And we still
  56. 1:52eat bananas.
  57. 1:53>> You said there was a 10 to 20% chance of
  58. 1:55killer robots wiping out humanity. So,
  59. 1:58>> you sounded worried then. Now you seem
  60. 2:00very relaxed about it. Have you changed
  61. 2:02your view?
  62. 2:03>> I still think there's risk
  63. 2:05associated with AI and robots. It's not
  64. 2:07zero.
  65. 2:08Um I I guess I've come to
  66. 2:12my my I sort of philosophical conclusion
  67. 2:15is to look on the bright side.
  68. 2:17Um
  69. 2:19I like I I can't see any way to really
  70. 2:21stop this
  71. 2:23incredible momentum of AI and robots.
  72. 2:26Um and
  73. 2:27and at times I sort of think, well,
  74. 2:29perhaps, you know, we're even if there
  75. 2:31was a stop button, we probably shouldn't
  76. 2:33press it
  77. 2:34uh because
  78. 2:37I cuz the most likely outcome
  79. 2:40is
  80. 2:41incredible
  81. 2:42abundance for
  82. 2:44for all.
  83. 2:45>> That's what I thought, actually. It
  84. 2:46struck me that now you seem to have
  85. 2:48decided it's inevitable.
  86. 2:51Let's hope for the best and go for the
  87. 2:53ride.
  88. 2:54>> Yeah. I mean
  89. 2:56yes, pretty much.
  90. 2:58Let's uh
  91. 2:59let's enjoy the ride as as
  92. 3:01I think my philosophy at this point. Cuz
  93. 3:03I mean, honestly, this this
  94. 3:04it's it's there seems to be just an
  95. 3:06inexorable
  96. 3:08uh progress here in in AI and and in
  97. 3:12robots
  98. 3:13that um
  99. 3:15>> But you
  100. 3:15>> you know, even if I wanted to stop it, I
  101. 3:18I don't I couldn't.
  102. 3:19>> Do you think there's still a 10 to 20%
  103. 3:21chance that killer robots
  104. 3:24mean the end of humanity?
  105. 3:27That's a high bar to as they say.
  106. 3:30>> Well,
  107. 3:32in long run we're all dead, of course.
  108. 3:33Um you know, um
  109. 3:35So, I mean, I I I so
  110. 3:36>> I mean
  111. 3:37>> the
  112. 3:37let let's
  113. 3:38according to
  114. 3:40the standard model of physics, the end
  115. 3:42of the universe is um
  116. 3:44a long, boring
  117. 3:47sort of thin, gray soup of
  118. 3:51uh
  119. 3:52gradual cooling.
  120. 3:54Um I mean, it is sort of the heat death
  121. 3:56of the universe is called. So, now if if
  122. 3:58the heat death of the universe is in
  123. 4:00fact the outcome for our reality, then
  124. 4:03it really is all about the journey.
  125. 4:05Because the destination is terrible.
  126. 4:07>> years is not [laughter] very
  127. 4:08Yeah, yeah, I get that, but that could
  128. 4:09be a long way ahead.
  129. 4:10>> long very very very very long way.
  130. 4:12>> If it's 10 to 20% in 10 years, you would
  131. 4:15you go into one of your rockets if you
  132. 4:16thought there was a 10 to 20% chance of
  133. 4:18it blowing up and killing you?
  134. 4:21>> Yes, but I I would
  135. 4:22>> [laughter]
  136. 4:23>> But let's say there's you can't do
  137. 4:25anything about it.
  138. 4:26>> Okay.
  139. 4:27>> But I guess the only thing we can try to
  140. 4:29do is minimize the probability of the
  141. 4:31bad outcome.
  142. 4:33Um
  143. 4:34but trying to stop the intelligence in
  144. 4:36robots
  145. 4:38um
  146. 4:40I mean, for the longest time I
  147. 4:44declined to participate in AI or was,
  148. 4:47you know,
  149. 4:48you know, created OpenAI as
  150. 4:51essentially a counterweight to to Google
  151. 4:52cuz at the time they were
  152. 4:55had more or less a monopoly on AI. In
  153. 4:57creating OpenAI and then having
  154. 4:59Anthropic spin out of that and now
  155. 5:01currently Anthropic's leader in AI.
  156. 5:03Um
  157. 5:05I mean, these
  158. 5:07these actions have actually resulted in
  159. 5:10knock-on effects that accelerated AI
  160. 5:13which wasn't really my intention.
  161. 5:15So, it just seems like all roads lead to
  162. 5:19acceleration of AI.
  163. 5:22So, then I'm like, okay, well,
  164. 5:24you can just sort of be sad about it or
  165. 5:27join the club, I suppose.
  166. 5:29>> Well, you've certainly joined the club.
  167. 5:30>> Yeah.
  168. 5:30>> I mean, you've
  169. 5:31>> [laughter]
  170. 5:31>> you've definitely joined the club.
  171. 5:33>> If you can't beat them, join them, you
  172. 5:34know.
  173. 5:34>> A number of the other leaders of of
  174. 5:38model makers have been pushing for
  175. 5:40different ways to try and minimize the
  176. 5:42risks. Um you know, just last week Demis
  177. 5:44Hassabis came out with a proposal for a
  178. 5:47private public regulator. Do you think
  179. 5:50you guys should be working more actively
  180. 5:53to get something in place to control the
  181. 5:55risk, or do you think that's impossible,
  182. 5:57too?
  183. 5:57>> Yeah.
  184. 5:58My recommendation to Demis cuz we I
  185. 6:00talked to him for
  186. 6:01a few hours before he published his his
  187. 6:03piece was that um
  188. 6:06the most immediate thing that we could
  189. 6:07do is to have the leading AI companies
  190. 6:10um
  191. 6:11at least just meet uh or have a some
  192. 6:13sort of call
  193. 6:15once every few weeks
  194. 6:17um
  195. 6:17and just
  196. 6:19uh discuss
  197. 6:21say any safety and security issues.
  198. 6:23>> But why why haven't you done that yet? I
  199. 6:25mean, you don't all like each other very
  200. 6:27much. Yeah, but you don't you're not all
  201. 6:29I mean, you're also all competing,
  202. 6:31right? You all want to be first.
  203. 6:32>> but that's that's where I think um good
  204. 6:36the competitors can keep each
  205. 6:37keep each other honest. So, meaning
  206. 6:40meaning like if um like I think it's
  207. 6:42quite difficult for someone in the
  208. 6:44government who doesn't have a deep
  209. 6:45technical understanding and isn't
  210. 6:47driving the frontier of of AI to know
  211. 6:50whether something should be released or
  212. 6:51not.
  213. 6:53Um however, the competitors
  214. 6:56I think what can, you know, if given
  215. 6:58sort of a week or two to review a new
  216. 7:00model um can highlight issues and uh
  217. 7:04and have an incentive to keep the all
  218. 7:06all the competitors have an incentive to
  219. 7:07keep the others honest.
  220. 7:09>> So, you'd have a system where the model
  221. 7:10makers would check each other's work. If
  222. 7:12there was something that worried them,
  223. 7:13you'd tell the government.
  224. 7:14>> Yeah.
  225. 7:15>> The government would then take
  226. 7:16>> If if there's something else was
  227. 7:17worrying them and that company was not
  228. 7:18doing anything to address that risk,
  229. 7:20then that would be the
  230. 7:23the moment for government to step in
  231. 7:24and take action, which is literally what
  232. 7:26just what occurred with my model.
  233. 7:28>> this all have to happen?
  234. 7:31>> Well, I think it'd be good to do that.
  235. 7:33I think
  236. 7:35immediately, really.
  237. 7:37>> Yeah, that's what people keep telling
  238. 7:38me. We we have to get this some kind of
  239. 7:39system in the next 6 months because of
  240. 7:42the pace at which the models are
  241. 7:43improving.
  242. 7:45>> Yeah, 6 months is a long time.
  243. 7:47Um
  244. 7:49>> So, you think shorter than that?
  245. 7:51>> Well,
  246. 7:53I mean there's there's so many AI
  247. 7:55breakthrough announcements.
  248. 7:56Sometimes multiple per day.
  249. 7:59I mean, how many were there last week?
  250. 8:00>> A lot.
  251. 8:01>> I lost track, you know.
  252. 8:02>> Yeah, a lot.
  253. 8:04>> Um
  254. 8:06>> So, so you you basically, you know,
  255. 8:08whether it's Demis's proposal or others,
  256. 8:10you need to get together and get this
  257. 8:11set up kind of now. It sounds like
  258. 8:13you're telling me.
  259. 8:14>> Yeah, I I I would recommend that we at
  260. 8:17least have some sort of informal um
  261. 8:21you know,
  262. 8:21weekly or bi-weekly
  263. 8:23um
  264. 8:24call and
  265. 8:26and and that like so just that that
  266. 8:27there's that that
  267. 8:30um there's maybe
  268. 8:31a week or 2 weeks of of early access by
  269. 8:35competitors. And and this is why I think
  270. 8:37the incentives work is because if if,
  271. 8:39you know, competing companies I think
  272. 8:41are likely to
  273. 8:43understand if something is a security
  274. 8:44risk. And they're they're not going to
  275. 8:46be shy about highlighting that that
  276. 8:49their their competitors' models should
  277. 8:50be delayed in release.
  278. 8:52>> You guys don't much like each other and
  279. 8:54you don't much trust each other. You're
  280. 8:55insulting each other on on social media
  281. 8:57constantly. So, do you do you think you
  282. 9:00can do that with people that you dislike
  283. 9:02that that where there seems to be such
  284. 9:03kind of personal vitriol between some of
  285. 9:05these leaders?
  286. 9:07>> Um
  287. 9:09Yes, well, I'm not I'm not a fan of Sam
  288. 9:11Altman because if you started a
  289. 9:12nonprofit that was meant to be an
  290. 9:14open-source
  291. 9:16AI company
  292. 9:18and um
  293. 9:20you know,
  294. 9:20owned by the world and it somehow got
  295. 9:22turned into an $800 billion for-profit
  296. 9:25company
  297. 9:26with closed source, I think you'd be
  298. 9:27like, "Well, wait a second. That's
  299. 9:30the exact opposite of what I donated the
  300. 9:31money for." That that's my issue. I
  301. 9:34think it's a legitimate one.
  302. 9:36Um
  303. 9:37And then Dario I think also is not a fan
  304. 9:39of Sam Altman because he felt in fact
  305. 9:42the reason the Anthropic team left
  306. 9:44OpenAI is because they didn't trust Sam
  307. 9:45Altman.
  308. 9:47Otherwise Anthropic
  309. 9:48wouldn't exist. They would still be at
  310. 9:49OpenAI.
  311. 9:51Um
  312. 9:52I I think Dario is a very principled
  313. 9:55person and he and he cares about things
  314. 9:56a lot.
  315. 9:57Uh he cares about the future of of the
  316. 9:59world and I I think everyone I've met at
  317. 10:02Anthropic so far has been
  318. 10:04I'm on Like no one set off my evil
  319. 10:06detector. You know, they always seem
  320. 10:07seem like they're well-intentioned.
  321. 10:10Um
  322. 10:11and uh
  323. 10:12and the road to hell is I think mostly
  324. 10:14paved with bad intentions. There are a
  325. 10:16few well-intentioned paving stones in
  326. 10:17there, so we we don't want to be
  327. 10:19complacent.
  328. 10:20Um
  329. 10:22but uh
  330. 10:23you know, at the end of the day if we
  331. 10:23have to talk, we'll talk. I mean
  332. 10:25set aside our personal differences for
  333. 10:27the good of the world type of thing.

About this transcript

This page contains the full transcript of Elon Musk on AI: humans will no longer be in control in ten years | The Economist by The Economist, generated from the public captions YouTube serves with the video. The transcript has 1,788 words across 333 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.