YouTube2Text

The Kind of Thinker Who Will Thrive in the AI Era | Head of Google Research (Yossi Matias) — Transcript

by Sinead Bovell · 4,314 words · 652 segments · language en · Watch on YouTube

Full transcript

  1. 0:00A lot of the magic happens actually when
  2. 0:01we bring together various people and but
  3. 0:03also connecting disciplines, connecting
  4. 0:05dots.
  5. 0:06>> What you described, that's a systems
  6. 0:07thinker.
  7. 0:08>> [music]
  8. 0:08>> And in some ways it's like an
  9. 0:09organization of one. That the person
  10. 0:11that you're hiring has to also think
  11. 0:13like a miniature company. You have all
  12. 0:15of these miniature departments. You're
  13. 0:16building, you're thinking about ethics,
  14. 0:18you're asking the right questions,
  15. 0:19[music] and then you're continually
  16. 0:20evolving like a company would. So what
  17. 0:22types of roles or skills do you think
  18. 0:26you would be hiring for in the future
  19. 0:28that maybe don't exist today? [music]
  20. 0:29>> We're looking for colleagues and workers
  21. 0:32that can actually ask the right
  22. 0:33question. That can actually then
  23. 0:34identify the opportunity. Sometime we
  24. 0:36call them soft skills, but they're
  25. 0:37skills, right? In the world where every
  26. 0:39researcher may have their own virtual
  27. 0:41lab, if you will, the researcher becomes
  28. 0:43what today we think about as advisers or
  29. 0:46researchers who have their own labs.
  30. 0:48>> I feel like I wouldn't do my job if I
  31. 0:49don't share a prediction with you. I
  32. 0:51have a feeling. So right now we have
  33. 0:52electrical engineers, but when something
  34. 0:54goes wrong in your house, you call an
  35. 0:56electrician. So if you think about this
  36. 0:58future with ambient intelligence and
  37. 1:00maybe there's AI that you can just talk
  38. 1:01to your wall. What is that role going to
  39. 1:03be? An AI-trician [music] where the
  40. 1:05person you call to fix the AI embedded
  41. 1:07in your house and your device, who's
  42. 1:09that going to be?
  43. 1:12I'm in Palo Alto right now at one of
  44. 1:13Google's headquarters. I just got off
  45. 1:15the stage with Yossi Matias. He's the VP
  46. 1:17of Google Research. And I describe
  47. 1:19Google Research as one of the places
  48. 1:21inside Google where the future gets
  49. 1:23prototyped before all of us get to
  50. 1:25experience it. So I wanted to ask him,
  51. 1:27what is he building? What are they
  52. 1:29working on? Where does he see all of
  53. 1:31this going
  54. 1:32>> [music]
  55. 1:32>> and what does that mean for all of us?
  56. 1:34I'm Shayne Bovell and this is I Have
  57. 1:36Questions.
  58. 1:39This episode is brought to you by Google
  59. 1:41Research.
  60. 1:42>> Thank you so much for having me. Good
  61. 1:43morning to to Yossi and to everyone.
  62. 1:46>> So I've um I've had the opportunity to
  63. 1:48to go behind the scenes of of Google
  64. 1:50Research these past few months. I've met
  65. 1:52with your your researchers and and your
  66. 1:54scientists. I've learned about what
  67. 1:55you're building, what's coming up, and
  68. 1:58and the scientist view on where the the
  69. 2:00future could be going.
  70. 2:02One thing is becoming clear, the role of
  71. 2:04a scientist could be changing. So, if
  72. 2:07we're if we're to fast forward a few
  73. 2:08years to 2030, what do you think the
  74. 2:10role of a scientist is? What does it
  75. 2:12look like?
  76. 2:14>> One thing which is I think quite
  77. 2:15exciting if you look into what why are
  78. 2:17we actually becoming scientists.
  79. 2:20It's really about curiosity, it's about
  80. 2:21exploring, it's about inventing. In
  81. 2:24fact, there's a key word. I remember
  82. 2:25there were these two series that I
  83. 2:26really loved reading. One was about
  84. 2:29explorers.
  85. 2:30One was about inventors.
  86. 2:32And to some extent, I think that what
  87. 2:35we're doing as researchers and
  88. 2:36scientists is in a way being inventors
  89. 2:38and explorers. But how we're doing it is
  90. 2:41changing all the time.
  91. 2:43Now, the world of exploration is just
  92. 2:45expanding, right? Before we had to do
  93. 2:48all this work in order to make this
  94. 2:50progress. And now suddenly, a lot of
  95. 2:52this heavy lifting we can actually use
  96. 2:54technology
  97. 2:56to actually help us with that. Then
  98. 2:58suddenly, we need to think differently.
  99. 2:59Now, for example, even today,
  100. 3:02in the world where every researcher may
  101. 3:04have their own virtual lab, if you will.
  102. 3:06And not only virtual lab, it's
  103. 3:08essentially an AI collaborator that is a
  104. 3:10polymath because it reads across
  105. 3:11disciplines.
  106. 3:13So, suddenly, in a way the the
  107. 3:16researcher becomes what today we think
  108. 3:18about as advisors or researchers who
  109. 3:20have their own labs, which is more into
  110. 3:23asking the right questions, into doing
  111. 3:25these iterations. So, in a way,
  112. 3:28this is up-leveling everything that we
  113. 3:30do. So, I I do believe that in the
  114. 3:32future, actually, the role of scientists
  115. 3:33is going to be more similar to what we
  116. 3:36do today as when you think about
  117. 3:38advisors, about senior
  118. 3:40senior scientists that are actually
  119. 3:42working with, you know, labs of grad
  120. 3:44students, postdocs, and so forth. So,
  121. 3:46it's in a way going to evolve into that.
  122. 3:49And Um, pace of that is something that
  123. 3:51we need to pay attention to, because
  124. 3:53we're not really set for that, right?
  125. 3:54The academics systems and what we're
  126. 3:56doing. So, I think this is time of
  127. 3:58transformation, which is um, you know,
  128. 4:01it uh, it can be quite intimidating
  129. 4:03sometime, but I think the opportunity
  130. 4:05there is uh, is is quite extraordinary.
  131. 4:08>> And it's really cool that you said you
  132. 4:10have this virtual lab, so this team of
  133. 4:11scientists that can follow you through
  134. 4:13your life. The lab goes with you. It's
  135. 4:15mobile in a way, I think which is really
  136. 4:18interesting and and what that concept
  137. 4:19looks like as a scientist moves
  138. 4:20throughout their life. Does their team
  139. 4:22of scientists, their AI agents move with
  140. 4:24them? Uh, I think that's something cool.
  141. 4:26>> And if I can add to that,
  142. 4:27I think that one of the things that um,
  143. 4:30I believe that a lot of the magic
  144. 4:31happens actually when we bring together
  145. 4:33various people, but also connecting
  146. 4:36disciplines, connecting dots. One thing
  147. 4:38that we observe over the years is that
  148. 4:40um, scientific work is driving people to
  149. 4:43more specialization, even within an
  150. 4:44area.
  151. 4:45And therefore becoming more difficult
  152. 4:48actually to work across disciplines. I
  153. 4:50mean, Hilbert, for example, the great
  154. 4:52mathematician actually started
  155. 4:54studying doing physics when
  156. 4:56pretty late in his career.
  157. 4:59And and Einstein actually published some
  158. 5:01of these papers with a with a colleague
  159. 5:03on who and you see uh, he was doing the
  160. 5:06physics and his colleague did the math.
  161. 5:09Now, again, the opportunity now to work
  162. 5:11with
  163. 5:12uh, systems that are can actually bring
  164. 5:14together those disciplines and enable
  165. 5:16everybody to actually work more um,
  166. 5:19across disciplines.
  167. 5:20And therefore actually look into things
  168. 5:23more um, in a holistic way. I think this
  169. 5:25is going to evolve over time, which I
  170. 5:27think is very exciting. I think still
  171. 5:29specialization and knowing to go deep is
  172. 5:31going to be incredibly important.
  173. 5:33>> Mhm. Yeah, I would agree.
  174. 5:35Uh, and if we were to to expand the
  175. 5:36picture outside of just the role of a
  176. 5:38scientist, how does my life change in in
  177. 5:412030? I've I've heard about what some of
  178. 5:42the researchers here are building. What
  179. 5:44does that look like for me in a few
  180. 5:46years?
  181. 5:46>> Yeah, it's interesting. So, we can ask
  182. 5:48ourselves actually
  183. 5:50um
  184. 5:51let me return the question to you.
  185. 5:53You're a futurist. So, uh
  186. 5:55are you are you looking more into the
  187. 5:57future and to the past as a futurist?
  188. 5:59And
  189. 5:59>> It's probably split. Probably.
  190. 6:01>> Okay, so if you would ask the same
  191. 6:03question say about uh how is how did our
  192. 6:06life change versus 10 years ago? What
  193. 6:08would be our answers?
  194. 6:10>> Interesting. Maybe the role of con-
  195. 6:13continuous communication. I think it's
  196. 6:16something that's transformed our world
  197. 6:18in a way that we don't even think about,
  198. 6:20that we're in contact with people all
  199. 6:21the time. Everywhere we go, the world
  200. 6:23comes with us.
  201. 6:25I think is maybe a way that it's
  202. 6:26changed.
  203. 6:27>> Yeah, I think in Okay, so so I think
  204. 6:29that uh in some ways things are changing
  205. 6:32dramatically and in other ways they are
  206. 6:34not, right? Because some of the things
  207. 6:35that I mean, what do we do? We look
  208. 6:37after interesting problems or important
  209. 6:39problems for our lives and how we
  210. 6:41communicate with each other. Obviously,
  211. 6:44uh with technology we actually enable
  212. 6:46building bridges. So, I think that it's
  213. 6:49amazing that today we can have
  214. 6:50connectivity among friends and families
  215. 6:52and so forth. At the same time, we're
  216. 6:55dealing with uh more information coming
  217. 6:57our way, so we need to manage that. So,
  218. 6:59I think we're going to see the same in
  219. 7:01many aspects of our life and work where
  220. 7:03suddenly we have all these tools and
  221. 7:05capabilities.
  222. 7:06And in the way there are um one one
  223. 7:09observation about technology is that
  224. 7:11once we start using it and it's good
  225. 7:13enough, we actually stop paying
  226. 7:15attention to that. We just use it. And
  227. 7:18uh and that's uh I I like to call it as
  228. 7:20ambient intelligence. So, the best form
  229. 7:22of intelligence that we're using around
  230. 7:24us is that it's working so good that we
  231. 7:26just use it. We don't actually pay
  232. 7:28attention to that. And then we're still
  233. 7:30trying to achieve our goals. So, I think
  234. 7:32that we're going to see this
  235. 7:34acceleration in many areas. You know, it
  236. 7:36was just a few years ago actually that
  237. 7:37we're talking about when and some people
  238. 7:40actually were asking if there are going
  239. 7:42to be self-driving cars. And now, if you
  240. 7:44go outside of this building and around
  241. 7:47in this area, you see so many Waymo
  242. 7:49cars. And this is, by the way, one
  243. 7:51unique experience that even though for
  244. 7:53most technologies after I use them a few
  245. 7:55times, I'm getting used to it, it's a
  246. 7:57technology that is just working and I
  247. 7:59still find it magical every time I do.
  248. 8:02So.
  249. 8:02>> Yeah, it's true. And and I know you had
  250. 8:04mentioned a little bit about Earth AI,
  251. 8:05which we've talked a lot about and I've
  252. 8:08gotten to meet some of your your
  253. 8:09researchers working on it. And when you
  254. 8:11combine that with that that idea of
  255. 8:13ambient intelligence and the questions
  256. 8:15that you'd be able to ask Earth AI. You
  257. 8:16know, someone with a a business
  258. 8:18background, being able to ask you,
  259. 8:20"Where should Where should the shop go?"
  260. 8:22right? If you want to launch a business,
  261. 8:24where where should it be located? And if
  262. 8:25you're a nonprofit, that's a question
  263. 8:28that you wouldn't have been able been
  264. 8:29able to hire a team to figure that out
  265. 8:31of consultants.
  266. 8:32But you can just ask Earth AI that
  267. 8:34question, you know, "Do I make this shop
  268. 8:35or should I lease this building?" These
  269. 8:37are really extraordinary questions.
  270. 8:39>> Yeah, I think it's a great example
  271. 8:40because it can if you go enough
  272. 8:42backwards, there the question is even
  273. 8:45for say responders or governments,
  274. 8:47"Where is the flood currently
  275. 8:48happening?" was out of reach. At some
  276. 8:50point it becomes available through
  277. 8:51satellite imagery and so forth, but when
  278. 8:53I started looking into that, this was 10
  279. 8:55years ago, actually visiting India and
  280. 8:57seeing actually a village that was
  281. 8:58flooded overnight,
  282. 9:00the question of where's where is the
  283. 9:02flood going to hit tomorrow was totally
  284. 9:04out of reach. It was totally a surprise.
  285. 9:07And now, actually, we have pretty good
  286. 9:09predictions, so people can actually
  287. 9:10anticipate where it's going to be and
  288. 9:12take action ahead of time. Uh in fact,
  289. 9:15just a few months ago, the government of
  290. 9:17Nigeria sent millions of dollars to
  291. 9:19villagers um ahead of a flood. They were
  292. 9:23using our our API for what we call Flood
  293. 9:25Hub, where we actually put our
  294. 9:26prediction capabilities. So, in a way,
  295. 9:29they could actually see the future, take
  296. 9:31action, and actually keep people safe.
  297. 9:33Now, with Earth AI, we can take it to
  298. 9:35the next level because today, when
  299. 9:37you're taking these predictions some
  300. 9:38people need to ask okay so what actions
  301. 9:40should we do what priorities where are
  302. 9:42people located but just imagine you can
  303. 9:44just ask a question so in this upcoming
  304. 9:46flood or in upcoming storm
  305. 9:49what are the impacted communities what
  306. 9:50is the infrastructure that is likely
  307. 9:52going to be impacted how can actually
  308. 9:54mobilize in for help so that to help
  309. 9:56people similarly think about how to do
  310. 9:59the same in health care how to convert
  311. 10:01the signals that we're getting from our
  312. 10:03devices into actionable information such
  313. 10:05as hey you may want to call a doctor so
  314. 10:07it's really about the object the
  315. 10:09objectives are not changing what is
  316. 10:12changing is the level in which we can
  317. 10:14actually expect to actually be able to
  318. 10:17operate on them on our daily life on our
  319. 10:19work life on humanity challenges and so
  320. 10:21forth so it's all the time up leveling
  321. 10:24and again if you go back I remember when
  322. 10:28you know the days that people ask me so
  323. 10:31what's what's going to what are kids
  324. 10:32going to do if you if they can just
  325. 10:35Google every fact or look it in
  326. 10:37Wikipedia right I mean are they going to
  327. 10:39be lazy or stupid well I think they're
  328. 10:40much smarter right because that's a
  329. 10:42given now and now we can actually take
  330. 10:44to the next level so I think we're going
  331. 10:46to see up leveling everything that we do
  332. 10:48but we may not notice it because it's
  333. 10:50just going to happen organically
  334. 10:52>> using ambient AI I had a vision of
  335. 10:54getting a text from your city official
  336. 10:55we've transferred a hundred dollars to
  337. 10:56your bank account we anticipate a flood
  338. 10:58coming to your neighborhood we're going
  339. 10:59to need you and your family to evacuate
  340. 11:01and then we just don't even think about
  341. 11:02that and that just becomes natural so in
  342. 11:04that very that future of ambient
  343. 11:06intelligence where we're just streaming
  344. 11:08it we don't think about it and it just
  345. 11:09works what would need to be true from a
  346. 11:11capability standpoint in AI for that to
  347. 11:14be the case what what does it need to
  348. 11:16enable
  349. 11:17>> I think it's already happening because
  350. 11:19it's coming all the time in a way that
  351. 11:21we just assume that it works so um you
  352. 11:24know that that's why we're working on
  353. 11:26all this foundational work all the
  354. 11:27foundational work on understanding the
  355. 11:30AI and new models so that actually these
  356. 11:33capabilities once they're getting into a
  357. 11:36certain threshold of trustworthiness and
  358. 11:39effectiveness and helpfulness, we just
  359. 11:41embed it in in our life and we just take
  360. 11:43them for granted. Uh by the way, this
  361. 11:45reminds me that talking reflecting back
  362. 11:47I remember when my son Michael who I
  363. 11:50think is somewhere here uh was in school
  364. 11:54uh
  365. 11:54he was came to me one day and said, "Hey
  366. 11:56dad, what's going on with your translate
  367. 11:58thing? I just actually translated this
  368. 12:00page and one line was actually not very
  369. 12:02good." And I thought to myself just a
  370. 12:04few years earlier this was actually
  371. 12:05science fiction to even have a
  372. 12:07translation. In fact, you had to send it
  373. 12:09to translators to do the work for you
  374. 12:11and now this became kind of in a way
  375. 12:13assumed that you just do it right. So, I
  376. 12:16think the same applies for practically
  377. 12:18everything that we do. A lot of stuff
  378. 12:20that today we expect um a lot of hard
  379. 12:22work or experts etc. is going to be
  380. 12:25facilitated with technology. We're just
  381. 12:26going to use it and then ask the next
  382. 12:28question.
  383. 12:30I expect us to be as busy as we are
  384. 12:31today because just our goals and targets
  385. 12:34are going to uh we're going to raise the
  386. 12:36bar of what we're trying to do.
  387. 12:38>> One thing that's interesting, I'm really
  388. 12:40curious to to to see how many people
  389. 12:42would consider themselves a data
  390. 12:43scientist at Google uh because that term
  391. 12:45that role wasn't invented until 2008
  392. 12:48after the data revolution from the last
  393. 12:50general purpose technology, the
  394. 12:51internet.
  395. 12:52So, what types of roles or skills do you
  396. 12:56think you would be hiring for in the
  397. 12:57future that maybe don't exist today?
  398. 13:01>> Yeah, the question for all is quite
  399. 13:02interesting because if I try again to
  400. 13:03reflect on roles today and how do they
  401. 13:05compare to what we expected before, I
  402. 13:07think they're quite different. Um
  403. 13:10I think that the opportunity for
  404. 13:12builders is going to be huge, but what
  405. 13:14does it mean to be a builder is changing
  406. 13:15all the time because of the knowledge,
  407. 13:17the tools, the capabilities to do that.
  408. 13:20You how to build, but also how to do
  409. 13:21research is going to be important and
  410. 13:23more important than ever. So, I would
  411. 13:24advise
  412. 13:25uh you know, how you do that these are
  413. 13:27part of the skills, but it's in a way
  414. 13:28also part of the role. Because what are
  415. 13:30we actually looking for when we are
  416. 13:33working with others? We're looking for
  417. 13:35colleagues and workers that can actually
  418. 13:37ask the right questions. That uh that
  419. 13:40can actually then identify the
  420. 13:41opportunities, that can use technologies
  421. 13:43to the best possible way to actually
  422. 13:45build things that matter. Now, what's
  423. 13:48also
  424. 13:49going to be more important than ever is
  425. 13:51the is the how, is the capabilities that
  426. 13:54sometimes we call them soft skills, but
  427. 13:56they're skills, right? It's how to
  428. 13:57communicate, how to listen. Because
  429. 14:00really, if you want to ask the right
  430. 14:01questions, you need to understand what
  431. 14:03matters. You need to apply judgment. Uh
  432. 14:05and and you also to apply the right
  433. 14:08judgment on the ethics of how to do it
  434. 14:10in the right way, how to be sensitive to
  435. 14:13what are the trade-offs. So, I think all
  436. 14:15these if if I try to then translate it
  437. 14:18to what we really we should expect and
  438. 14:20how we should actually want the next
  439. 14:22generation to grow into, we want them to
  440. 14:24be skillful to ask questions, to be able
  441. 14:26to do research, to use technologies, and
  442. 14:29importantly to know how to adapt and how
  443. 14:32to learn, because the pace of change is
  444. 14:34such that in any given moment what we're
  445. 14:35doing is going to change. Now,
  446. 14:38we also should recognize that we're now
  447. 14:40in kind of a transition because there
  448. 14:42are all these very powerful technologies
  449. 14:44and there's a lot of chatter about oh, I
  450. 14:46can just use it to solve problems, but
  451. 14:49there's a certain friction in using
  452. 14:50them. But, I think we should assume that
  453. 14:52much of the friction that we
  454. 14:54uh see around us is actually going to go
  455. 14:56away because somebody is going to help
  456. 14:59actually collectively we're going to
  457. 15:01reduce this, and therefore what we're
  458. 15:03going to remain with are these
  459. 15:05empowerment and use it. So, it's really
  460. 15:08I don't think we're going to be in lack
  461. 15:10of problems that we need to solve. And
  462. 15:13how we're going to solve them is going
  463. 15:14to change all the time. So, being able
  464. 15:16to adapt to that is going to be quite
  465. 15:18important.
  466. 15:19>> And what you described uh you know, for
  467. 15:21for the skills of the future or who you
  468. 15:23could be hiring, they have to think
  469. 15:25about ethics, what questions are they
  470. 15:26asking, they also have to be in somewhat
  471. 15:28a builder.
  472. 15:30So, that's a systems thinker and in some
  473. 15:32ways it's like an organization of one
  474. 15:34like that the person that you're hiring
  475. 15:35has to also think like a miniature
  476. 15:37company. You have all of these miniature
  477. 15:39departments, you're building, you're
  478. 15:40thinking about ethics, you're asking the
  479. 15:42right questions, and then you're
  480. 15:43continually evolving like a company
  481. 15:45would, which is also a really
  482. 15:46interesting way to to think about it.
  483. 15:49>> Yeah, in a way it's related to earlier
  484. 15:51uh we talked about scientists and I was
  485. 15:53thinking about how AI can actually help
  486. 15:55us be across more disciplines. So, you
  487. 15:57don't necessarily need to find another
  488. 15:59expert to be able to do some progress.
  489. 16:01You can actually use technology in order
  490. 16:03to actually work more across
  491. 16:05disciplines. Similarly, in other domains
  492. 16:07actually now when I hear about uh you
  493. 16:11know, people with no experience with
  494. 16:12coding using live coding to actually do
  495. 16:14basic uh coding and therefore enable
  496. 16:18empowers them to actually do more with
  497. 16:20that. And similarly, when people can
  498. 16:22actually uh sketch out some uh you know,
  499. 16:25product design and so forth. So, this
  500. 16:27notion of everyone can play roles in
  501. 16:30different ways by using technology,
  502. 16:32again, I think it's going to accelerate
  503. 16:34and this is incredibly important. By the
  504. 16:37way, it will always we looked into for
  505. 16:39example uh product leaders uh as uh
  506. 16:43people that would have uh computer
  507. 16:46science degree or engineering degree or
  508. 16:47something quite important to know how to
  509. 16:50make those connections, how to build
  510. 16:51those bridges. So, a lot of what we do
  511. 16:53is really bringing together things and I
  512. 16:55think that to your point, the ability
  513. 16:57for every person to actually play more
  514. 17:00roles in doing that is very empowering.
  515. 17:03As is by the way, the ability to ask
  516. 17:05questions and get analysis and
  517. 17:07practically get uh kind of a if you
  518. 17:09will, a small seminar work done on your
  519. 17:13behalf with an with a with a system that
  520. 17:16can give you a very quick acceleration
  521. 17:18of taking decisions of
  522. 17:21helping you to also decide what you need
  523. 17:23to build as well as how to build it.
  524. 17:26>> And I think I feel like I wouldn't do my
  525. 17:27job if I didn't share a prediction with
  526. 17:30you. I have a feeling so right now we
  527. 17:31have electrical engineers, but when
  528. 17:33something goes wrong in your house, you
  529. 17:35call an electrician. So if you think
  530. 17:37about this future with ambient
  531. 17:38intelligence and maybe there's AI that
  532. 17:39you can just talk to your wall, what is
  533. 17:41that role going to be an AI-trician
  534. 17:43where the person you call to fix the AI
  535. 17:46embedded in your house and your device?
  536. 17:48Who's that going to be? I I wonder what
  537. 17:50what that world's going to look like and
  538. 17:51I have a feeling we'll be hiring for
  539. 17:52that.
  540. 17:53>> I'm sure many here are getting calls
  541. 17:55from family to fix the Wi-Fi at home. So
  542. 17:57>> Yeah.
  543. 17:58>> Hopefully hopefully we can solve that
  544. 18:00problem at some point.
  545. 18:01>> think I work in IT, so
  546. 18:04>> There you go.
  547. 18:06>> [laughter]
  548. 18:06>> So if we're to to move to the present
  549. 18:10for a second with all of this, it feels
  550. 18:12pretty cool what's in the pipeline, what
  551. 18:14will be possible, the problems that
  552. 18:16we'll be able to solve, the systems
  553. 18:17thinkers that we're training. What would
  554. 18:20be keeping you up at night in this
  555. 18:21moment?
  556. 18:24>> Uh
  557. 18:25>> [snorts]
  558. 18:25>> you know, it's
  559. 18:27the the one thing that I think is is
  560. 18:29really because the opportunity is so
  561. 18:31vast and so broad and practically
  562. 18:35everything that we look around there's
  563. 18:36something we can do it to make it
  564. 18:38better.
  565. 18:39I think that the biggest thing on my
  566. 18:40mind is how do we make sure that we're
  567. 18:42really meeting the moment and taking the
  568. 18:44right opportunities to work on and to
  569. 18:46tackle.
  570. 18:47There's huge variability on what we can
  571. 18:49do. We can use now technology and you
  572. 18:51know, we have an amazing talent that
  573. 18:54every project, every effort, everything
  574. 18:56that we do can make a difference. The
  575. 18:58question is how do we do the biggest
  576. 19:00difference that we can? And again, if I
  577. 19:02think about the opportunity that we have
  578. 19:04today to take
  579. 19:06societal problems, anything from health
  580. 19:08care to crisis resilience to education
  581. 19:11to energy development.
  582. 19:14In all of those, there's an opportunity
  583. 19:16to actually to have an opportunity that
  584. 19:18can make impact on so many people in
  585. 19:21such a profound way.
  586. 19:23And making the right choices of how to
  587. 19:25actually go after the right opportunity
  588. 19:28is to me one thing that I think
  589. 19:29everybody should worry about or think
  590. 19:31about because of the urgency.
  591. 19:34And again, I'm incredibly optimistic
  592. 19:36about how we can actually use innovation
  593. 19:38and technology and working together in
  594. 19:40order to actually address and
  595. 19:43give a positive impact on on people's
  596. 19:46lives, daily life, business, but also
  597. 19:49importantly on all the societal
  598. 19:50challenges that we have. Think about
  599. 19:52health care.
  600. 19:53That's huge opportunity for us to
  601. 19:55actually save lives, improve lives, and
  602. 19:57we're already seeing good progress on
  603. 19:59that. So, that's actually the main
  604. 20:01question on my mind is how do I actually
  605. 20:03get the fastest and the most effective
  606. 20:05way?
  607. 20:06And also, because there are so many
  608. 20:08opportunities when I as when I think
  609. 20:11about the Google research organization
  610. 20:13and what we're doing and the teams, the
  611. 20:15amazing talent that we have, one
  612. 20:17question that we're constantly asking,
  613. 20:19on the one hand, we can have a lot of
  614. 20:21impact here and now and very soon. On
  615. 20:24the other hand, actually an important
  616. 20:25mission that we have is really to look
  617. 20:28into what's on the horizon, what's
  618. 20:29around the corner, what are the new
  619. 20:31innovation or things that nobody was
  620. 20:33thinking about yet that are going to
  621. 20:35make big impact into the future. A lot
  622. 20:38of the things that currently were
  623. 20:40working on, actually, were not were not
  624. 20:42on the radar just a few years ago or
  625. 20:44sometimes a few months ago. So, what are
  626. 20:46the next things that we can uncover, we
  627. 20:48can actually make the next big jumps in
  628. 20:50every area that we have? I think the
  629. 20:52opportunity is exciting and is there,
  630. 20:55and it's something, by the way, that we
  631. 20:56encourage. We actually spent here a few
  632. 20:58days now asking what are the big
  633. 21:00questions that we need to ask in all
  634. 21:02levels, and importantly, how can we
  635. 21:04bring everybody in collaboration so as
  636. 21:07to we can focus on the next big
  637. 21:09questions and have the biggest possible
  638. 21:10impact.
  639. 21:11>> I think
  640. 21:13the bigger the questions become and the
  641. 21:14more advanced technologies that we get
  642. 21:16to to solve these problems with, I think
  643. 21:18we're all going to be up later. I think
  644. 21:20that that's what the future looks like
  645. 21:21too. Everybody working all night with
  646. 21:23their agents. Yoav, see it's been a
  647. 21:25pleasure. Thank you all so much.
  648. 21:26>> Thank you so much.
  649. 21:28Thanks, Renee.
  650. 21:29>> Thank you all.
  651. 21:29>> [applause]
  652. 21:30>> All right.

About this transcript

This page contains the full transcript of The Kind of Thinker Who Will Thrive in the AI Era | Head of Google Research (Yossi Matias) by Sinead Bovell, generated from the public captions YouTube serves with the video. The transcript has 4,314 words across 652 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 translate the transcript into 100+ languages.

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, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.