1 00:00:00,458 --> 00:00:01,584 Most US jobs, 2 00:00:01,584 --> 00:00:02,502 the threat to your job, 3 00:00:02,502 --> 00:00:03,169 it’s not AI. 4 00:00:03,628 --> 00:00:04,546 It’s some other person 5 00:00:04,546 --> 00:00:06,423 who uses AI better than you do. 6 00:00:11,803 --> 00:00:12,595 Welcome to WorkLab, 7 00:00:12,595 --> 00:00:13,847 the podcast from Microsoft. 8 00:00:14,139 --> 00:00:15,223 I’m your host, Molly Wood. 9 00:00:15,473 --> 00:00:16,224 On WorkLab, 10 00:00:16,224 --> 00:00:17,767 we talk to experts about AI 11 00:00:17,767 --> 00:00:19,144 and the future of work. 12 00:00:19,519 --> 00:00:20,812 We talk about how technology 13 00:00:20,812 --> 00:00:21,688 can supercharge 14 00:00:21,688 --> 00:00:22,981 human capabilities, 15 00:00:22,981 --> 00:00:25,525 how it can transform entire organizations 16 00:00:25,525 --> 00:00:26,735 and turn them into 17 00:00:26,735 --> 00:00:28,611 what we like to call Frontier Firms. 18 00:00:28,862 --> 00:00:30,655 Today, we’re joined by Tyler Cowen. 19 00:00:30,655 --> 00:00:32,282 Tyler is the Holbert L. Harris 20 00:00:32,282 --> 00:00:33,199 Chair of Economics 21 00:00:33,199 --> 00:00:34,701 at George Mason University, 22 00:00:34,909 --> 00:00:36,661 author and deeply influential 23 00:00:36,661 --> 00:00:38,329 public thinker who co-writes 24 00:00:38,329 --> 00:00:39,748 the Marginal Revolution blog, 25 00:00:40,040 --> 00:00:40,665 and hosts the 26 00:00:40,665 --> 00:00:42,834 ‘Conversations with Tyler’ podcast. 27 00:00:43,168 --> 00:00:44,252 Today, he’s going to endure 28 00:00:44,252 --> 00:00:45,420 being interviewed 29 00:00:45,420 --> 00:00:47,005 on the other side of the microphone. 30 00:00:47,005 --> 00:00:47,881 In the past few years, 31 00:00:47,881 --> 00:00:48,673 he has been tracking 32 00:00:48,673 --> 00:00:49,883 how generative AI 33 00:00:49,883 --> 00:00:51,384 is impacting the economy, 34 00:00:51,384 --> 00:00:53,470 labor markets, and institutions. 35 00:00:53,720 --> 00:00:54,554 His writing explores 36 00:00:54,554 --> 00:00:56,014 how technology is redefining 37 00:00:56,014 --> 00:00:57,640 productivity, creativity, 38 00:00:57,640 --> 00:00:58,850 and the skills that will matter 39 00:00:58,850 --> 00:00:59,809 most in the years ahead. 40 00:01:00,101 --> 00:01:01,728 Tyler, thanks for joining me on WorkLab. 41 00:01:02,020 --> 00:01:03,438 Molly, happy to be here. 42 00:01:03,772 --> 00:01:06,149 Okay, so I want to ask you first about 43 00:01:06,691 --> 00:01:08,818 kind of a shift from you, actually, 44 00:01:08,818 --> 00:01:11,279 your 2011 book, “The Great Stagnation” 45 00:01:11,279 --> 00:01:12,697 got you this kind of reputation 46 00:01:12,697 --> 00:01:15,825 as a skeptic, a technological pessimist. 47 00:01:16,201 --> 00:01:17,077 You characterize the US 48 00:01:17,077 --> 00:01:19,412 as being stuck in an innovative slump, 49 00:01:19,412 --> 00:01:21,164 and today 50 00:01:21,164 --> 00:01:21,831 you’re a little bit 51 00:01:21,831 --> 00:01:23,666 more of an energetic optimist about tech, 52 00:01:23,666 --> 00:01:24,375 I would say. 53 00:01:24,584 --> 00:01:25,668 Do you think that’s fair? 54 00:01:25,668 --> 00:01:26,419 And if so, 55 00:01:26,419 --> 00:01:27,670 what flipped that switch for you? 56 00:01:28,046 --> 00:01:28,671 That’s fair. 57 00:01:28,671 --> 00:01:30,381 But keep in mind, in my 2011 58 00:01:30,381 --> 00:01:31,925 book, “The Great Stagnation,” 59 00:01:31,925 --> 00:01:32,842 I predicted we 60 00:01:32,842 --> 00:01:33,343 would get out of 61 00:01:33,343 --> 00:01:34,302 the great stagnation 62 00:01:34,302 --> 00:01:35,804 in the next 20 years. 63 00:01:35,804 --> 00:01:37,972 And my book after that, “Average is Over,” 64 00:01:37,972 --> 00:01:39,349 I think that’s 2013, 65 00:01:39,766 --> 00:01:41,059 was saying AI is what’s 66 00:01:41,059 --> 00:01:42,227 going to pull us out. 67 00:01:42,227 --> 00:01:43,186 So I think my view’s 68 00:01:43,186 --> 00:01:44,687 been pretty consistent 69 00:01:44,687 --> 00:01:46,523 and I’m proud of the predictions 70 00:01:46,523 --> 00:01:47,857 in the earlier book. 71 00:01:47,857 --> 00:01:49,150 And we’re finally at the point 72 00:01:49,150 --> 00:01:50,151 where it’s happening. 73 00:01:50,151 --> 00:01:53,154 I would say biomedical advances and AI, 74 00:01:53,154 --> 00:01:55,490 which are now also meshing together, 75 00:01:55,490 --> 00:01:57,408 mean we’re back to real progress again. 76 00:01:57,408 --> 00:01:58,493 It’s fantastic. 77 00:01:58,493 --> 00:01:59,369 The funny thing is, 78 00:01:59,369 --> 00:02:00,703 people don’t always like it. 79 00:02:00,703 --> 00:02:02,038 They actually prefer 80 00:02:02,038 --> 00:02:03,456 the world of stagnation. 81 00:02:03,873 --> 00:02:04,958 Ooh, say more about that. 82 00:02:05,500 --> 00:02:07,168 It’s more predictable. 83 00:02:07,168 --> 00:02:08,169 In any given year, 84 00:02:08,169 --> 00:02:09,963 your income may not go up as much, 85 00:02:09,963 --> 00:02:10,880 but you know what your job 86 00:02:10,880 --> 00:02:11,548 will look like 87 00:02:11,548 --> 00:02:13,299 for the next 10 or 20 years, 88 00:02:13,550 --> 00:02:15,218 and that’s comforting to people. 89 00:02:15,343 --> 00:02:16,427 If I think of my own job, 90 00:02:16,427 --> 00:02:18,096 university professor, 91 00:02:18,096 --> 00:02:19,389 I started thinking about it 92 00:02:19,389 --> 00:02:20,557 when I was 14. 93 00:02:20,557 --> 00:02:21,933 Now I’m 63. 94 00:02:21,933 --> 00:02:23,810 That’s almost 50 years. 95 00:02:23,810 --> 00:02:25,687 What I do has changed a bit, 96 00:02:25,687 --> 00:02:26,771 but it’s the same job 97 00:02:26,771 --> 00:02:28,189 I thought I was getting. 98 00:02:28,189 --> 00:02:29,566 And that will not be true 99 00:02:29,566 --> 00:02:31,693 for many people in the coming generation. 100 00:02:31,943 --> 00:02:33,069 Right. That is a fair point. 101 00:02:33,069 --> 00:02:34,445 We are going to talk a fair bit 102 00:02:34,445 --> 00:02:35,613 about change management 103 00:02:35,613 --> 00:02:36,656 because as you point out, 104 00:02:36,656 --> 00:02:39,075 we are in a moment of extreme change. 105 00:02:39,534 --> 00:02:40,743 What did you see then that made 106 00:02:40,743 --> 00:02:41,619 you think AI would be 107 00:02:41,619 --> 00:02:42,954 the thing that pulled us out? 108 00:02:42,954 --> 00:02:44,622 And what do you think 109 00:02:44,622 --> 00:02:45,748 about where we are now? 110 00:02:46,040 --> 00:02:47,250 When I was very young, 111 00:02:47,250 --> 00:02:48,459 I was a chess player, 112 00:02:48,459 --> 00:02:49,502 and I saw chess-playing 113 00:02:49,502 --> 00:02:51,546 computers back in the 1970s. 114 00:02:51,546 --> 00:02:52,714 And they were terrible. 115 00:02:52,714 --> 00:02:53,590 But I saw the pace 116 00:02:53,590 --> 00:02:54,841 at which they advanced. 117 00:02:54,841 --> 00:02:56,634 And by 1997, 118 00:02:56,634 --> 00:02:59,053 you have Deep Blue beating Garry Kasparov 119 00:02:59,053 --> 00:03:00,430 and being the strongest 120 00:03:00,430 --> 00:03:02,307 chess-playing entity in the world. 121 00:03:02,599 --> 00:03:04,017 What chess players know is 122 00:03:04,017 --> 00:03:05,518 how much intuition is in chess. 123 00:03:05,518 --> 00:03:07,228 It’s not mainly calculation. 124 00:03:07,228 --> 00:03:08,605 So if you’ve seen computers 125 00:03:08,605 --> 00:03:09,856 advance in chess, 126 00:03:09,856 --> 00:03:11,941 you know they can handle intuition, 127 00:03:11,941 --> 00:03:12,692 which is something 128 00:03:12,692 --> 00:03:14,652 a lot of people at the time didn’t see. 129 00:03:14,652 --> 00:03:15,528 But it’s interesting 130 00:03:15,528 --> 00:03:17,488 that myself, Kasparov, 131 00:03:17,488 --> 00:03:18,781 who is a chess player, 132 00:03:18,781 --> 00:03:20,283 and Ken Rogoff, an economist 133 00:03:20,283 --> 00:03:22,202 and also a champion chess player, 134 00:03:22,202 --> 00:03:23,453 we all were very early 135 00:03:23,453 --> 00:03:24,871 to be bullish on AI, 136 00:03:24,871 --> 00:03:26,247 and it’s all for the same reason. 137 00:03:26,247 --> 00:03:27,123 The link to chess. 138 00:03:27,582 --> 00:03:28,625 Well, 139 00:03:28,625 --> 00:03:30,084 I mean, that is a conversation 140 00:03:30,084 --> 00:03:32,212 I would like to have for an hour, 141 00:03:32,212 --> 00:03:32,712 but won’t. 142 00:03:33,379 --> 00:03:34,464 Let’s fast forward to 143 00:03:34,464 --> 00:03:37,425 roughly today or around 2022 144 00:03:37,425 --> 00:03:40,220 when certainly AI had existed 145 00:03:40,220 --> 00:03:41,012 for a long time before, 146 00:03:41,012 --> 00:03:42,263 but generative AI 147 00:03:42,263 --> 00:03:43,765 then has created 148 00:03:43,765 --> 00:03:45,433 this sort of new tsunami. 149 00:03:46,100 --> 00:03:47,227 How would you describe 150 00:03:47,227 --> 00:03:49,270 the economic impact 151 00:03:49,812 --> 00:03:51,689 now, since its introduction? 152 00:03:52,106 --> 00:03:53,858 Well, right now it’s still very small. 153 00:03:53,858 --> 00:03:54,984 If you’re a programmer. 154 00:03:54,984 --> 00:03:56,694 It’s had a big impact on you. 155 00:03:56,694 --> 00:03:57,862 If you’re good at using it, 156 00:03:57,862 --> 00:03:59,364 it does a lot of your work. 157 00:03:59,364 --> 00:04:00,907 It makes it harder for some 158 00:04:00,949 --> 00:04:02,492 maybe junior-level programmers 159 00:04:02,492 --> 00:04:03,451 to get hired. 160 00:04:03,826 --> 00:04:04,827 Most of the economy, 161 00:04:04,827 --> 00:04:06,120 I would say it saves people 162 00:04:06,120 --> 00:04:07,121 a lot of time. 163 00:04:07,121 --> 00:04:08,706 It’s treated as an add on. 164 00:04:08,706 --> 00:04:10,416 It’s turned into leisure work 165 00:04:10,416 --> 00:04:12,085 that might have taken you three hours. 166 00:04:12,085 --> 00:04:13,628 It’s done in five minutes. 167 00:04:13,628 --> 00:04:15,255 But in terms of GDP growth, 168 00:04:15,255 --> 00:04:16,506 productivity growth, 169 00:04:16,506 --> 00:04:17,548 we’re not there yet. 170 00:04:17,882 --> 00:04:20,134 Let’s stay on this this line of 171 00:04:20,134 --> 00:04:21,344 of where we are and where we’re going. 172 00:04:21,344 --> 00:04:22,428 Because certainly 173 00:04:22,428 --> 00:04:23,763 it is the way of tech leaders 174 00:04:23,763 --> 00:04:24,555 to predict 175 00:04:24,555 --> 00:04:25,682 an immediate 176 00:04:25,682 --> 00:04:26,808 productivity boom 177 00:04:26,808 --> 00:04:28,518 from any new technology. 178 00:04:28,643 --> 00:04:31,604 Some economic experts predicted the same. 179 00:04:31,688 --> 00:04:32,939 Do you think the impact 180 00:04:32,939 --> 00:04:35,900 is, in fact, slower than 181 00:04:35,900 --> 00:04:36,651 predicted, 182 00:04:36,651 --> 00:04:38,903 or was this also part of the chess game? 183 00:04:38,903 --> 00:04:39,112 or was this also part of the chess game? 184 00:04:39,112 --> 00:04:41,030 It depends who’s doing the prediction, 185 00:04:41,030 --> 00:04:41,906 but I think we need 186 00:04:41,906 --> 00:04:43,992 a whole new generation of firms 187 00:04:43,992 --> 00:04:45,451 and institutions. 188 00:04:45,451 --> 00:04:47,620 If you take currently existing companies 189 00:04:47,620 --> 00:04:49,706 or nonprofits or universities 190 00:04:49,706 --> 00:04:50,540 and try to get them 191 00:04:50,540 --> 00:04:53,084 to reorganize themselves around AI, 192 00:04:53,084 --> 00:04:54,502 mostly you will fail. 193 00:04:54,711 --> 00:04:56,629 I mean, think back to the earlier years 194 00:04:56,629 --> 00:04:58,381 of the American auto industry. 195 00:04:58,381 --> 00:04:59,757 Toyota is a kind of threat 196 00:04:59,757 --> 00:05:00,925 to General Motors. 197 00:05:00,925 --> 00:05:02,302 Try to get General Motors 198 00:05:02,302 --> 00:05:04,429 to do the good things Toyota was doing. 199 00:05:04,429 --> 00:05:06,806 That’s arguably a much simpler problem, 200 00:05:06,806 --> 00:05:08,308 but mostly they failed. 201 00:05:08,308 --> 00:05:10,101 And we just didn’t adjust in time. 202 00:05:10,101 --> 00:05:11,686 So adjusting to AI, 203 00:05:11,686 --> 00:05:14,522 which is a much more radical change, 204 00:05:14,522 --> 00:05:17,275 you need startups, which will do things 205 00:05:17,275 --> 00:05:18,735 like supply legal services 206 00:05:18,735 --> 00:05:21,112 or medical diagnosis or whatever. 207 00:05:21,112 --> 00:05:22,071 And over time 208 00:05:22,071 --> 00:05:23,031 they will cycle through 209 00:05:23,031 --> 00:05:25,700 and generationally become dominant firms. 210 00:05:25,908 --> 00:05:27,618 But that could take 20 years or more. 211 00:05:27,618 --> 00:05:28,870 It will happen bit by bit, 212 00:05:28,870 --> 00:05:31,164 piece by piece, step by step. 213 00:05:31,164 --> 00:05:31,372 Yeah. 214 00:05:31,372 --> 00:05:32,206 This is something you’ve written 215 00:05:32,206 --> 00:05:32,540 about that. 216 00:05:32,540 --> 00:05:34,208 Institutions specifically 217 00:05:34,208 --> 00:05:35,543 these, you know, universities, 218 00:05:35,543 --> 00:05:36,461 big businesses, 219 00:05:36,461 --> 00:05:37,420 governments 220 00:05:37,420 --> 00:05:40,340 are the real bottleneck to AI progress. 221 00:05:40,423 --> 00:05:41,799 Do you think that it is possible 222 00:05:41,799 --> 00:05:43,468 for a legacy institution 223 00:05:43,468 --> 00:05:44,761 to reinvent itself 224 00:05:44,761 --> 00:05:45,553 as a Frontier Firm, 225 00:05:45,553 --> 00:05:46,763 or do we have to wait for this startup 226 00:05:46,763 --> 00:05:48,056 cycle to play out? 227 00:05:48,556 --> 00:05:50,308 I think we mostly have to wait. 228 00:05:50,308 --> 00:05:51,642 I think we’ll be surprised 229 00:05:51,642 --> 00:05:53,436 that some companies will do it. 230 00:05:53,436 --> 00:05:54,479 They may be companies 231 00:05:54,479 --> 00:05:55,813 which are founder-led, 232 00:05:55,813 --> 00:05:57,732 where the founder can storm into the room 233 00:05:57,732 --> 00:05:59,233 and just say, “We’re going to do this,” 234 00:05:59,233 --> 00:06:01,694 and no one can challenge that logic. 235 00:06:01,944 --> 00:06:03,446 But it’s just not going to be 236 00:06:03,446 --> 00:06:04,864 that many players in the game. 237 00:06:04,864 --> 00:06:07,450 So I think it will be painful. 238 00:06:07,450 --> 00:06:09,660 Hardest of all, with nonprofits 239 00:06:09,660 --> 00:06:11,162 and especially universities 240 00:06:11,162 --> 00:06:12,080 and governments, 241 00:06:12,538 --> 00:06:16,042 the IRS still gets millions of faxes 242 00:06:16,042 --> 00:06:18,336 every year and uses fax machines. 243 00:06:18,336 --> 00:06:19,670 So you’re telling the IRS, 244 00:06:19,670 --> 00:06:21,422 “Organize everything around AI.” 245 00:06:21,756 --> 00:06:23,424 How about telling the IRS, 246 00:06:23,424 --> 00:06:24,592 “Use the internet more.” 247 00:06:24,592 --> 00:06:26,803 We haven’t even quite managed to do that. 248 00:06:26,803 --> 00:06:28,096 So yes, it will be slow. 249 00:06:28,346 --> 00:06:28,638 Yeah. 250 00:06:28,930 --> 00:06:30,431 It seems like everybody sort of 251 00:06:30,431 --> 00:06:31,599 feels like we’re in a bubble 252 00:06:31,599 --> 00:06:32,475 and we’re headed for a crash, 253 00:06:32,475 --> 00:06:33,935 and it’s going to come any minute now. 254 00:06:35,103 --> 00:06:36,312 Do you agree? 255 00:06:36,604 --> 00:06:37,480 I would make this point. 256 00:06:37,480 --> 00:06:38,064 Right now 257 00:06:38,064 --> 00:06:39,524 tech sector earnings 258 00:06:39,524 --> 00:06:41,067 are higher than tech sector 259 00:06:41,067 --> 00:06:42,819 CapEx expenditures. 260 00:06:42,819 --> 00:06:43,319 That’s good. 261 00:06:43,319 --> 00:06:44,278 It means they can afford 262 00:06:44,278 --> 00:06:45,571 to spend the money. 263 00:06:45,571 --> 00:06:46,781 Do I think every big 264 00:06:46,781 --> 00:06:48,574 or even small tech company 265 00:06:48,658 --> 00:06:50,076 will earn more because of this? 266 00:06:50,076 --> 00:06:50,326 No. 267 00:06:50,326 --> 00:06:51,119 I think a lot of them 268 00:06:51,119 --> 00:06:52,286 will lose a lot of money. 269 00:06:52,412 --> 00:06:53,413 It’s like the railroads. 270 00:06:53,413 --> 00:06:55,248 We did overbuild the railroads, 271 00:06:55,248 --> 00:06:56,874 but the railroads didn’t go away. 272 00:06:56,874 --> 00:06:58,835 In fact, they transformed the world. 273 00:06:58,835 --> 00:07:00,378 So it’s not like Pets.com, 274 00:07:00,378 --> 00:07:01,421 where one day it’s gone 275 00:07:01,421 --> 00:07:03,506 and you say, “Boo hoo. It was a bubble.” 276 00:07:03,506 --> 00:07:05,675 AI is here to last. It’s going to work. 277 00:07:05,675 --> 00:07:06,676 There’ll be major winners. 278 00:07:06,676 --> 00:07:08,177 There’ll be a lot of losers. 279 00:07:08,177 --> 00:07:09,679 A very common story. 280 00:07:09,679 --> 00:07:10,805 If you want to call that a bubble 281 00:07:10,805 --> 00:07:11,848 because there are losers, 282 00:07:11,848 --> 00:07:12,849 that’s fine, 283 00:07:12,849 --> 00:07:14,183 but it’s not a bubble in the sense 284 00:07:14,183 --> 00:07:16,185 that when a bubble pops, it’s gone. 285 00:07:17,270 --> 00:07:17,895 I think you could argue 286 00:07:17,895 --> 00:07:18,855 that the parallel, actually, 287 00:07:18,855 --> 00:07:20,440 the Pets.com parallel, 288 00:07:20,815 --> 00:07:22,275 might apply now, 289 00:07:22,275 --> 00:07:23,526 which is that the internet 290 00:07:23,526 --> 00:07:24,360 didn’t go anywhere. 291 00:07:24,360 --> 00:07:24,861 You know the 292 00:07:24,861 --> 00:07:26,446 this new economy didn’t go anywhere. 293 00:07:26,446 --> 00:07:29,031 We just had a little bit of a speed bump. 294 00:07:29,031 --> 00:07:30,074 And we do in fact buy 295 00:07:30,074 --> 00:07:31,492 our dog food at home online. 296 00:07:31,492 --> 00:07:32,618 And it’s delivered. 297 00:07:32,618 --> 00:07:32,869 Yeah. 298 00:07:33,786 --> 00:07:34,996 Not from Pets.com. 299 00:07:34,996 --> 00:07:36,414 Some other company, yeah. 300 00:07:36,747 --> 00:07:37,957 In an interview earlier this year, 301 00:07:37,957 --> 00:07:39,625 you suggested that colleges 302 00:07:39,625 --> 00:07:41,419 should dedicate at least a third 303 00:07:41,669 --> 00:07:43,713 of their curricula to AI literacy. 304 00:07:44,255 --> 00:07:45,631 Say more about how education 305 00:07:45,631 --> 00:07:46,507 needs to evolve 306 00:07:46,507 --> 00:07:48,176 to provide value in this era. 307 00:07:48,593 --> 00:07:49,886 To be clear, right now, 308 00:07:49,886 --> 00:07:51,345 we do not have the professors 309 00:07:51,345 --> 00:07:52,972 to teach AI literacy. 310 00:07:52,972 --> 00:07:54,849 So my suggestion is hypothetical. 311 00:07:54,849 --> 00:07:57,143 It’s not something we actually can do. 312 00:07:57,143 --> 00:07:59,645 But over the course of the next 20 years, 313 00:07:59,645 --> 00:08:01,355 basically almost every job 314 00:08:01,355 --> 00:08:03,232 I think even gardener, carpenter, 315 00:08:03,232 --> 00:08:04,358 physical jobs. 316 00:08:04,442 --> 00:08:05,485 If you’re in sports, 317 00:08:05,485 --> 00:08:08,029 San Antonio Spurs are claiming they’re now 318 00:08:08,029 --> 00:08:11,032 the AI-ready NBA basketball team. 319 00:08:11,073 --> 00:08:13,493 So if all jobs will require AI, 320 00:08:13,493 --> 00:08:14,535 we need to be teaching 321 00:08:14,535 --> 00:08:17,121 and training everyone in AI skills. 322 00:08:17,121 --> 00:08:17,830 It’s quite hard 323 00:08:17,830 --> 00:08:20,166 because AI now is changing so rapidly. 324 00:08:20,166 --> 00:08:21,918 You could learn something six months ago 325 00:08:21,918 --> 00:08:23,669 and then you’ve got to learn a new thing. 326 00:08:23,669 --> 00:08:25,213 So it will take a lot of time, 327 00:08:25,213 --> 00:08:26,422 and we need to teach people 328 00:08:26,422 --> 00:08:27,673 how to stay current. 329 00:08:27,673 --> 00:08:29,926 That itself changes. 330 00:08:29,926 --> 00:08:32,386 But what we need to do, in my opinion, 331 00:08:32,386 --> 00:08:34,096 is have writing 332 00:08:34,096 --> 00:08:34,847 be prominent 333 00:08:34,847 --> 00:08:36,974 in regular education at all levels, 334 00:08:36,974 --> 00:08:38,017 and I mean in-person, 335 00:08:38,017 --> 00:08:39,477 face-to-face writing, 336 00:08:39,477 --> 00:08:40,603 not take home homework 337 00:08:40,603 --> 00:08:42,146 where GPT does it for you, 338 00:08:42,772 --> 00:08:45,733 writing, numeracy, and AI. 339 00:08:45,858 --> 00:08:47,652 Those are the things we need to learn. 340 00:08:47,652 --> 00:08:48,319 The other things 341 00:08:48,319 --> 00:08:50,488 you can mostly learn from AI. 342 00:08:50,488 --> 00:08:51,280 If you can deal with 343 00:08:51,280 --> 00:08:52,823 numbers mentally, right? 344 00:08:52,907 --> 00:08:54,367 That means you can think well 345 00:08:54,408 --> 00:08:55,993 and work with the AIs, 346 00:08:55,993 --> 00:08:57,245 you’re in great shape. 347 00:08:57,370 --> 00:08:58,913 We need to have further training 348 00:08:58,913 --> 00:09:00,748 in all kinds of specialized skills. 349 00:09:00,748 --> 00:09:02,250 Say you want to be a chemist. 350 00:09:02,250 --> 00:09:02,917 Well, what do you need to 351 00:09:02,917 --> 00:09:04,168 learn about the lab? 352 00:09:04,168 --> 00:09:06,045 But you start off by emphasizing 353 00:09:06,045 --> 00:09:06,921 those three things: 354 00:09:06,921 --> 00:09:09,382 AI, writing, thinking, and math. 355 00:09:10,925 --> 00:09:11,592 You know, we talked a 356 00:09:11,592 --> 00:09:12,552 lot about Frontier Firms, 357 00:09:12,552 --> 00:09:13,219 but you’re talking about 358 00:09:13,219 --> 00:09:14,679 Frontier universities in a way. 359 00:09:14,679 --> 00:09:15,596 If you’re saying that there are no 360 00:09:15,596 --> 00:09:17,056 there’s simply are not there now. 361 00:09:17,056 --> 00:09:17,723 Exactly. 362 00:09:18,140 --> 00:09:19,850 Arizona State is maybe 363 00:09:19,850 --> 00:09:21,352 the most forward-thinking. 364 00:09:21,978 --> 00:09:22,812 I have some hopes 365 00:09:22,812 --> 00:09:24,230 for University of Austin. 366 00:09:24,230 --> 00:09:25,982 I think they will be trying. 367 00:09:26,482 --> 00:09:28,276 But, you know, personnel is policy, 368 00:09:28,276 --> 00:09:29,318 as they say. 369 00:09:29,318 --> 00:09:30,319 And in universities, 370 00:09:30,319 --> 00:09:32,196 people feel threatened by AI. 371 00:09:32,196 --> 00:09:34,031 They know it will change their jobs. 372 00:09:34,031 --> 00:09:35,992 It doesn’t have to make them worse. 373 00:09:35,992 --> 00:09:37,201 But I see very, 374 00:09:37,201 --> 00:09:37,827 very slow 375 00:09:37,827 --> 00:09:39,787 speeds of adaptation and adjustment. 376 00:09:40,997 --> 00:09:41,914 Adoption is just 377 00:09:41,914 --> 00:09:43,457 such a fundamental question in adoption, 378 00:09:43,457 --> 00:09:45,084 and I think the word you just used, 379 00:09:45,084 --> 00:09:46,586 which is adaptation. 380 00:09:46,669 --> 00:09:48,713 There is this kind of mindset shift, 381 00:09:48,754 --> 00:09:50,214 habit shift, skills shift, 382 00:09:50,214 --> 00:09:51,340 even within workplaces 383 00:09:51,340 --> 00:09:53,509 that already have some 384 00:09:53,509 --> 00:09:55,511 AI collaborators or digital employees. 385 00:09:55,511 --> 00:09:56,804 Is that fair to say? 386 00:09:56,804 --> 00:09:58,055 Absolutely fair to say. 387 00:09:58,055 --> 00:09:58,848 It’s hard 388 00:09:59,140 --> 00:10:00,766 for just about any institution 389 00:10:00,766 --> 00:10:01,976 to make the adjustment, 390 00:10:01,976 --> 00:10:03,978 including major AI companies. 391 00:10:03,978 --> 00:10:05,146 They don’t do most of their own 392 00:10:05,146 --> 00:10:07,565 internal tasks by AI, right? 393 00:10:07,565 --> 00:10:08,316 You would think, well, 394 00:10:08,316 --> 00:10:09,650 it should be easy for them. 395 00:10:09,650 --> 00:10:10,484 And, you know, 396 00:10:10,484 --> 00:10:11,986 I think they’ll manage over time, 397 00:10:11,986 --> 00:10:13,279 but it’s not easy for anyone. 398 00:10:14,030 --> 00:10:14,905 You’re saying, 399 00:10:14,905 --> 00:10:16,032 and there have also been 400 00:10:16,032 --> 00:10:17,617 a number of studies saying gen AI 401 00:10:17,617 --> 00:10:19,994 has had minimal impact 402 00:10:19,994 --> 00:10:21,495 on the labor market, 403 00:10:21,495 --> 00:10:22,997 but we are also seeing headlines 404 00:10:22,997 --> 00:10:25,124 about large-scale job cuts 405 00:10:25,124 --> 00:10:26,083 that are being framed 406 00:10:26,083 --> 00:10:28,377 as the result of embracing AI. 407 00:10:28,377 --> 00:10:30,296 Do you buy that framing, 408 00:10:30,296 --> 00:10:33,090 or is there a large corporate disconnect? 409 00:10:33,299 --> 00:10:35,301 I’m not persuaded by the framing. 410 00:10:35,301 --> 00:10:36,719 There could be some truth to it. 411 00:10:36,719 --> 00:10:38,846 I don’t think we know how much yet. 412 00:10:38,846 --> 00:10:39,972 I think there’s a good chance 413 00:10:39,972 --> 00:10:41,557 we’re headed for a recession 414 00:10:41,724 --> 00:10:43,684 for non-AI related reasons. 415 00:10:44,060 --> 00:10:45,895 And then of course, people get laid off. 416 00:10:46,145 --> 00:10:48,105 AI is an easy thing to blame. 417 00:10:48,105 --> 00:10:50,191 Oh, it’s not me, it’s the AI. 418 00:10:50,441 --> 00:10:52,485 What I suspect will happen 419 00:10:52,485 --> 00:10:54,654 is as AI becomes more important, 420 00:10:54,654 --> 00:10:56,447 new hirings will slow down. 421 00:10:56,447 --> 00:10:57,823 I’m not sure that many people 422 00:10:57,823 --> 00:10:59,950 would be fired in the US. 423 00:10:59,950 --> 00:11:01,827 If it’s a call center in the Philippines 424 00:11:01,827 --> 00:11:03,245 or in Hyderabad, India 425 00:11:03,412 --> 00:11:04,205 there, I can see 426 00:11:04,205 --> 00:11:05,122 people being laid off 427 00:11:05,122 --> 00:11:07,249 in the early stages because of AI. 428 00:11:07,249 --> 00:11:08,584 But most US jobs, 429 00:11:08,584 --> 00:11:09,543 the threat to your job, 430 00:11:09,543 --> 00:11:10,419 it’s not AI. 431 00:11:10,711 --> 00:11:11,587 It’s some of the person 432 00:11:11,587 --> 00:11:13,547 who uses AI better than you do. 433 00:11:13,547 --> 00:11:14,465 And there’s going to be a lot of 434 00:11:14,465 --> 00:11:15,716 job turnover due to that. 435 00:11:16,258 --> 00:11:18,761 Do we risk making AI into the boogeyman? 436 00:11:19,220 --> 00:11:21,013 We’re making AI into the boogeyman 437 00:11:21,013 --> 00:11:21,514 right now. 438 00:11:21,514 --> 00:11:22,431 Just go to Twitter 439 00:11:22,431 --> 00:11:23,933 and you see people upset 440 00:11:23,933 --> 00:11:25,518 at all sorts of people, companies, 441 00:11:25,518 --> 00:11:26,560 AI systems. 442 00:11:26,560 --> 00:11:27,395 My goodness, 443 00:11:27,395 --> 00:11:28,479 they’re so negative 444 00:11:28,479 --> 00:11:29,689 about one of humanity’s 445 00:11:29,689 --> 00:11:31,399 most marvelous inventions. 446 00:11:32,608 --> 00:11:33,984 Is what they’re saying on Twitter, 447 00:11:33,984 --> 00:11:34,735 or your response? 448 00:11:34,735 --> 00:11:35,778 What they’re saying on Twitter 449 00:11:35,778 --> 00:11:36,779 is so negative. 450 00:11:37,071 --> 00:11:38,989 There’s so many villains in the story, 451 00:11:38,989 --> 00:11:40,491 and we’ve done something amazing, 452 00:11:40,491 --> 00:11:41,617 but that’s common. 453 00:11:41,826 --> 00:11:43,327 People actually do enjoy stagnation 454 00:11:43,327 --> 00:11:44,912 and change is very scary. 455 00:11:44,912 --> 00:11:45,538 That’s right. 456 00:11:45,955 --> 00:11:48,666 And you’re talking about a realignment of 457 00:11:48,666 --> 00:11:50,000 business, education, 458 00:11:50,000 --> 00:11:51,252 and culture 459 00:11:51,252 --> 00:11:53,796 to try to integrate this technology 460 00:11:54,004 --> 00:11:55,256 like is it that is it that disruptive 461 00:11:55,381 --> 00:11:58,050 that that’s what we have to consider? 462 00:11:58,050 --> 00:11:58,968 I believe so. 463 00:11:58,968 --> 00:12:00,261 And you use the word scary. 464 00:12:00,261 --> 00:12:00,928 That’s correct. 465 00:12:00,928 --> 00:12:03,639 I would also say it can be demoralizing. 466 00:12:03,639 --> 00:12:06,642 So if I ask the best AI systems questions 467 00:12:06,642 --> 00:12:07,852 about economics, 468 00:12:07,852 --> 00:12:09,103 usually their answers 469 00:12:09,103 --> 00:12:10,730 are better than my answers. 470 00:12:10,730 --> 00:12:12,148 Now how should I feel about that? 471 00:12:13,149 --> 00:12:14,066 I would say I’m more 472 00:12:14,066 --> 00:12:15,943 delighted than upset, 473 00:12:15,943 --> 00:12:16,861 but I understand 474 00:12:16,861 --> 00:12:17,820 when people don’t like it. 475 00:12:17,820 --> 00:12:19,029 I’m like, “Whoa. 476 00:12:19,029 --> 00:12:20,906 What am I good for now?” 477 00:12:20,906 --> 00:12:22,616 And I have answers to that question, 478 00:12:22,616 --> 00:12:24,577 but it requires adjustment from me. 479 00:12:24,952 --> 00:12:25,619 Right. 480 00:12:26,036 --> 00:12:28,038 So then now put yourself into the context 481 00:12:28,038 --> 00:12:29,790 of somebody at work 482 00:12:29,790 --> 00:12:30,916 who is experiencing this. 483 00:12:31,208 --> 00:12:33,377 How do you manage for 484 00:12:33,377 --> 00:12:35,880 those potential feelings of inadequacy 485 00:12:35,880 --> 00:12:37,757 or the change in 486 00:12:38,382 --> 00:12:39,884 job style and behavior 487 00:12:40,217 --> 00:12:43,345 that will be necessary to adapt? 488 00:12:44,138 --> 00:12:45,765 I don’t think we know yet. 489 00:12:45,765 --> 00:12:46,599 This is a reason 490 00:12:46,599 --> 00:12:48,058 why I’m bullish on startups 491 00:12:48,058 --> 00:12:48,934 more generally. 492 00:12:48,934 --> 00:12:49,935 If you have a startup 493 00:12:49,935 --> 00:12:50,853 and you tell everyone 494 00:12:50,853 --> 00:12:51,854 you’re hiring, 495 00:12:51,854 --> 00:12:53,105 “Well, you need to know AI, 496 00:12:53,105 --> 00:12:55,024 we’re going to run things around AI,” 497 00:12:55,024 --> 00:12:57,193 the morale problem is much lower. 498 00:12:57,401 --> 00:12:58,861 Tobi Lütke looked at at Shopify, 499 00:12:58,861 --> 00:13:00,988 tried to do this with this famous memo. 500 00:13:00,988 --> 00:13:03,115 He said, “Anyone we hire at Shopify, 501 00:13:03,115 --> 00:13:03,949 from here on out, 502 00:13:03,949 --> 00:13:06,035 has to be very good with AI.” 503 00:13:06,035 --> 00:13:07,328 That was an excellent move. 504 00:13:07,703 --> 00:13:08,329 But it’s hard to 505 00:13:08,329 --> 00:13:09,288 see that through 506 00:13:09,288 --> 00:13:10,790 and then actually reward the people 507 00:13:10,790 --> 00:13:12,082 who do a good job of it. 508 00:13:12,082 --> 00:13:13,334 When a lot of your people, 509 00:13:13,417 --> 00:13:14,710 they’re probably using 510 00:13:14,710 --> 00:13:16,128 something like ChatGPT 511 00:13:16,128 --> 00:13:17,421 in their daily lives, 512 00:13:17,421 --> 00:13:18,631 but they haven’t really 513 00:13:18,839 --> 00:13:20,132 adjusted to the fact 514 00:13:20,132 --> 00:13:21,217 that it will dominate 515 00:13:21,217 --> 00:13:22,802 a lot of their workday. 516 00:13:23,260 --> 00:13:24,845 Let’s talk about sectors for a minute. 517 00:13:25,179 --> 00:13:26,806 What are the sectors 518 00:13:26,806 --> 00:13:28,265 you think in addition to 519 00:13:28,265 --> 00:13:29,225 maybe the types of work 520 00:13:29,225 --> 00:13:30,559 that are most likely to feel AI’s effects 521 00:13:30,559 --> 00:13:32,102 that are most likely to feel AI’s effects 522 00:13:32,102 --> 00:13:33,312 most intensely in the next, 523 00:13:33,312 --> 00:13:34,480 let’s say, year to 18 months? 524 00:13:35,439 --> 00:13:36,565 Year to 18 months. 525 00:13:36,565 --> 00:13:37,900 I think it will take 526 00:13:37,900 --> 00:13:39,485 a few more years 527 00:13:39,485 --> 00:13:41,487 than that to really matter. 528 00:13:41,487 --> 00:13:43,072 So there’s plenty of places 529 00:13:43,072 --> 00:13:44,240 like, say, law firms 530 00:13:44,240 --> 00:13:46,992 where AI is already super powerful, 531 00:13:46,992 --> 00:13:49,453 but you’re not allowed to send your query 532 00:13:49,453 --> 00:13:52,331 to someone’s foundation model out there 533 00:13:52,331 --> 00:13:53,541 in the Bay Area 534 00:13:53,541 --> 00:13:55,835 because it’s confidential information. 535 00:13:55,835 --> 00:13:56,961 So that will work 536 00:13:56,961 --> 00:13:57,419 when, say, 537 00:13:57,419 --> 00:13:59,338 a law firm on its own 538 00:13:59,338 --> 00:14:01,090 hard drive owns the whole model 539 00:14:01,090 --> 00:14:01,882 and controls it 540 00:14:01,882 --> 00:14:02,508 and doesn’t have to 541 00:14:02,508 --> 00:14:03,801 send the query and the answer 542 00:14:03,801 --> 00:14:04,718 anywhere. 543 00:14:04,927 --> 00:14:06,053 I don’t see that happening 544 00:14:06,053 --> 00:14:07,304 in the next 18 months. 545 00:14:07,304 --> 00:14:08,472 I can imagine 546 00:14:08,472 --> 00:14:10,516 it happening in the next three years. 547 00:14:10,850 --> 00:14:13,352 But right now, I think entertainment, 548 00:14:13,561 --> 00:14:15,437 the arts, graphic design, 549 00:14:15,437 --> 00:14:17,022 obviously programming. 550 00:14:17,356 --> 00:14:18,482 I think functions 551 00:14:18,482 --> 00:14:20,901 such as finance and customer service, 552 00:14:20,901 --> 00:14:22,862 they’re often pretty repetitive. 553 00:14:22,862 --> 00:14:23,362 That to me 554 00:14:23,362 --> 00:14:24,113 seems more like 555 00:14:24,113 --> 00:14:25,364 a two or three year thing. 556 00:14:25,364 --> 00:14:27,324 Not so much the next year. 557 00:14:27,324 --> 00:14:29,618 We just don’t quite seem to be there 558 00:14:29,618 --> 00:14:31,245 to have working agents 559 00:14:31,245 --> 00:14:33,122 that are sufficiently reliable 560 00:14:33,122 --> 00:14:33,914 that you would turn 561 00:14:33,914 --> 00:14:34,540 you know, the candy 562 00:14:34,540 --> 00:14:36,125 store over to them right now. 563 00:14:36,125 --> 00:14:37,167 It will come. 564 00:14:37,167 --> 00:14:38,377 I’m really quite sure. 565 00:14:38,377 --> 00:14:39,670 Less than five years, 566 00:14:39,670 --> 00:14:41,005 but not in the next year. 567 00:14:41,088 --> 00:14:42,464 Let’s talk about humans for a minute. 568 00:14:42,464 --> 00:14:43,340 In your recent writing, 569 00:14:43,340 --> 00:14:44,258 you have been exploring 570 00:14:44,258 --> 00:14:45,009 this sort of question 571 00:14:45,009 --> 00:14:46,010 about 572 00:14:46,010 --> 00:14:47,761 how AI might change 573 00:14:47,761 --> 00:14:49,305 what it means to be human. 574 00:14:49,305 --> 00:14:50,347 And we’ve talked about that 575 00:14:50,347 --> 00:14:52,474 a little bit in the context of learning. 576 00:14:52,474 --> 00:14:54,643 But what do you think humans 577 00:14:54,643 --> 00:14:58,022 most need to protect or preserve? 578 00:14:58,856 --> 00:14:59,940 You’ll need to build out 579 00:14:59,940 --> 00:15:01,859 your personal network much more. 580 00:15:01,859 --> 00:15:02,735 Everyone now 581 00:15:02,735 --> 00:15:04,695 has a perfectly written cover letter, 582 00:15:04,695 --> 00:15:06,322 so that does not distinguish you. 583 00:15:06,864 --> 00:15:07,740 Who can actually vouch 584 00:15:07,740 --> 00:15:08,282 for you, 585 00:15:08,282 --> 00:15:09,491 recommend you, 586 00:15:09,491 --> 00:15:10,951 speak to what you’ve done with them 587 00:15:10,951 --> 00:15:12,244 or for them. 588 00:15:12,244 --> 00:15:13,954 That was already super important. 589 00:15:13,954 --> 00:15:15,873 Now it’s much more important. 590 00:15:15,873 --> 00:15:16,999 I think your charisma, 591 00:15:16,999 --> 00:15:18,542 your physical presence. 592 00:15:18,542 --> 00:15:20,502 I hesitate to use the word looks 593 00:15:20,502 --> 00:15:22,004 because I don’t think simply 594 00:15:22,004 --> 00:15:23,005 being attractive looking 595 00:15:23,005 --> 00:15:25,007 is necessarily the main thing. 596 00:15:25,215 --> 00:15:26,800 But to have looks that fit 597 00:15:26,800 --> 00:15:27,843 what people think 598 00:15:27,843 --> 00:15:29,929 your looks ought to be like, 599 00:15:29,929 --> 00:15:32,348 and to be persuasive face-to-face, 600 00:15:32,681 --> 00:15:34,808 again, it’s always mattered a great deal. 601 00:15:34,808 --> 00:15:37,227 That’s just on the ascendency, 602 00:15:37,227 --> 00:15:40,189 the ability and willingness to travel, 603 00:15:40,189 --> 00:15:42,733 and some sort of interpersonal skill 604 00:15:42,733 --> 00:15:43,567 that reflects 605 00:15:43,567 --> 00:15:46,487 knowing how you can fit into a system 606 00:15:46,487 --> 00:15:48,489 and when you should defer to the AI, 607 00:15:48,489 --> 00:15:50,699 when you should not defer to the AI. 608 00:15:50,699 --> 00:15:51,867 And how can you persuade 609 00:15:51,867 --> 00:15:53,410 the people you’re working with 610 00:15:53,410 --> 00:15:55,621 to actually learn more about the AI? 611 00:15:55,621 --> 00:15:56,872 That’s a pretty tricky skill. 612 00:15:56,872 --> 00:15:58,958 We’re not used to testing for that, 613 00:15:58,958 --> 00:16:00,250 but we will need to test for it 614 00:16:00,250 --> 00:16:01,502 in the future. 615 00:16:01,502 --> 00:16:03,087 Do you mean literally test for it 616 00:16:03,087 --> 00:16:04,505 like an aptitue? 617 00:16:04,505 --> 00:16:05,255 Literally test for it. Yeah. 618 00:16:05,255 --> 00:16:06,840 I asked Sam Altman recently 619 00:16:06,840 --> 00:16:08,467 in my podcast with him, 620 00:16:08,467 --> 00:16:09,551 “Well, how do you make sure 621 00:16:09,551 --> 00:16:11,095 you’re hiring people who, 622 00:16:11,095 --> 00:16:11,887 when the time comes, 623 00:16:11,887 --> 00:16:13,889 are willing to step aside for AI 624 00:16:13,889 --> 00:16:14,515 and perform 625 00:16:14,515 --> 00:16:16,350 some other function in their jobs?” 626 00:16:16,350 --> 00:16:17,434 I don’t think any of us know 627 00:16:17,434 --> 00:16:18,352 how to do that yet. 628 00:16:18,352 --> 00:16:20,270 We’re used to asking other questions. 629 00:16:21,397 --> 00:16:22,272 I mean, certainly 630 00:16:22,272 --> 00:16:23,357 people aren’t good at doing it 631 00:16:23,565 --> 00:16:24,441 when a human comes along. 632 00:16:24,942 --> 00:16:25,609 That’s right. 633 00:16:25,609 --> 00:16:27,319 Maybe the AI is less threatening, right? 634 00:16:27,319 --> 00:16:28,946 Maybe you could sort of say, well, it is. 635 00:16:28,946 --> 00:16:29,780 It does have access 636 00:16:29,780 --> 00:16:30,990 to all the human knowledge in the world. 637 00:16:30,990 --> 00:16:32,074 So I guess 638 00:16:32,074 --> 00:16:33,617 it’s doing a better job than me. 639 00:16:33,826 --> 00:16:34,702 What will I mean, I guess on that note, 640 00:16:34,702 --> 00:16:37,496 then what will we need to let go of 641 00:16:37,496 --> 00:16:38,330 in addition to, 642 00:16:38,330 --> 00:16:39,540 it sounds like, a little bit of ego? 643 00:16:39,999 --> 00:16:41,792 I’m a university professor. 644 00:16:41,792 --> 00:16:44,211 I give students comments on their papers 645 00:16:44,211 --> 00:16:45,170 all the time. 646 00:16:45,170 --> 00:16:46,964 I now give them AI comments 647 00:16:46,964 --> 00:16:48,257 and my own comments. 648 00:16:48,257 --> 00:16:50,426 The AI comments are typically better 649 00:16:50,426 --> 00:16:51,677 than my comments, 650 00:16:51,677 --> 00:16:53,178 but my comments see things 651 00:16:53,178 --> 00:16:54,555 the AI doesn’t, 652 00:16:54,555 --> 00:16:56,140 like how the paper will play 653 00:16:56,140 --> 00:16:57,975 to certain human audiences, 654 00:16:57,975 --> 00:16:58,809 maybe what journal 655 00:16:58,809 --> 00:17:00,185 it should be published in. 656 00:17:00,644 --> 00:17:02,021 Something maybe temperamental 657 00:17:02,021 --> 00:17:02,771 about the paper 658 00:17:02,771 --> 00:17:04,732 that the AI is in a sense too 659 00:17:04,732 --> 00:17:06,525 focused on the substance, 660 00:17:06,525 --> 00:17:07,526 but on the substance, 661 00:17:07,526 --> 00:17:10,070 the AI just wipes the floor with me. 662 00:17:10,070 --> 00:17:13,032 So I need to invest more in learning 663 00:17:13,032 --> 00:17:14,450 these other skills 664 00:17:14,450 --> 00:17:16,535 and just skills of persuasion. 665 00:17:16,535 --> 00:17:18,287 How can I convince the students 666 00:17:18,287 --> 00:17:19,872 to take the AI seriously? 667 00:17:20,497 --> 00:17:21,040 Going back 668 00:17:21,040 --> 00:17:21,790 to sort of 669 00:17:21,790 --> 00:17:23,542 amplifying our essential humanness 670 00:17:23,542 --> 00:17:24,460 and making that maybe 671 00:17:24,460 --> 00:17:25,627 part of our education. 672 00:17:25,919 --> 00:17:26,962 I feel like when we say 673 00:17:26,962 --> 00:17:27,588 something like that, 674 00:17:27,588 --> 00:17:28,839 it feels like such a big lift, like, 675 00:17:28,839 --> 00:17:29,840 oh, how will we ever 676 00:17:30,049 --> 00:17:31,675 retool our education system 677 00:17:31,675 --> 00:17:32,926 to focus on humanity 678 00:17:32,926 --> 00:17:35,929 and the arts and interpersonal skills? 679 00:17:36,096 --> 00:17:37,598 But we really did used to have that. 680 00:17:37,598 --> 00:17:38,932 You know, it’s sort of fair 681 00:17:38,932 --> 00:17:42,186 to say we overemphasized on technology. 682 00:17:42,186 --> 00:17:42,811 And I mean, you’re 683 00:17:42,811 --> 00:17:43,771 an economics professor, 684 00:17:43,771 --> 00:17:46,065 but numbers are pure economics 685 00:17:46,065 --> 00:17:48,025 and data-driven decision making, 686 00:17:48,025 --> 00:17:50,652 and it feels doable to me 687 00:17:50,652 --> 00:17:51,904 that we could sort of return 688 00:17:51,904 --> 00:17:53,489 to a different educational model 689 00:17:53,489 --> 00:17:56,366 if we think the value will be there 690 00:17:56,366 --> 00:17:57,201 in the end. 691 00:17:57,451 --> 00:17:59,495 It’s not intrinsically hard to do at all. 692 00:17:59,495 --> 00:18:00,954 It’s a matter of will, 693 00:18:00,954 --> 00:18:02,122 and that the insiders 694 00:18:02,122 --> 00:18:04,708 in many institutions, schools included, 695 00:18:04,708 --> 00:18:06,710 they just don’t want to change very much. 696 00:18:06,710 --> 00:18:08,587 So those are hard nuts to crack, 697 00:18:08,587 --> 00:18:10,380 but just intrinsically, 698 00:18:10,380 --> 00:18:10,839 we’ve done it 699 00:18:10,839 --> 00:18:11,965 many times in the past, 700 00:18:11,965 --> 00:18:13,634 as you point out, and we’ll do it again. 701 00:18:13,634 --> 00:18:15,302 But like, how long and painful 702 00:18:15,302 --> 00:18:16,553 will the transition be? 703 00:18:16,553 --> 00:18:18,305 That’s what worries me. 704 00:18:18,305 --> 00:18:19,139 How long and painful 705 00:18:19,139 --> 00:18:20,474 do you think the transition will be? 706 00:18:21,433 --> 00:18:22,184 I don’t know, 707 00:18:22,184 --> 00:18:24,478 dragging everyone kicking and screaming. 708 00:18:24,478 --> 00:18:26,063 It’s decades, in my view. 709 00:18:26,063 --> 00:18:26,522 Yeah. 710 00:18:26,522 --> 00:18:27,606 But some places will adjust 711 00:18:27,606 --> 00:18:28,565 pretty quickly. 712 00:18:28,857 --> 00:18:30,275 Obviously we cannot prove a negative. 713 00:18:30,275 --> 00:18:31,235 We can’t prove an ROI 714 00:18:31,235 --> 00:18:32,778 that doesn’t yet exist. 715 00:18:32,778 --> 00:18:33,529 But it seems that 716 00:18:33,529 --> 00:18:35,239 the fundamental question, 717 00:18:35,239 --> 00:18:36,990 in terms of overcoming 718 00:18:36,990 --> 00:18:37,908 the kicking and screaming, 719 00:18:37,908 --> 00:18:38,742 if it were a toddler, 720 00:18:38,742 --> 00:18:39,535 you would say, 721 00:18:39,535 --> 00:18:41,703 “There’s a cookie at the end.” 722 00:18:41,703 --> 00:18:44,748 And I wonder how we attempt to measure 723 00:18:44,873 --> 00:18:47,876 the future value that makes this 724 00:18:47,918 --> 00:18:50,295 transition and this upheaval worthwhile? 725 00:18:50,671 --> 00:18:51,713 My personal view, 726 00:18:51,713 --> 00:18:53,799 which is speculative, to be clear, 727 00:18:53,799 --> 00:18:55,175 is that at some point 728 00:18:55,175 --> 00:18:56,301 younger people 729 00:18:56,301 --> 00:18:57,219 will basically not 730 00:18:57,219 --> 00:18:58,971 die of diseases anymore 731 00:18:58,971 --> 00:19:00,180 and will live to be 96 732 00:19:00,180 --> 00:19:02,933 or 97 years old and die of old age. 733 00:19:03,183 --> 00:19:04,434 That’s a huge cookie. 734 00:19:04,434 --> 00:19:05,477 Am I going to get that? 735 00:19:05,477 --> 00:19:06,812 I’m not counting on it. 736 00:19:06,812 --> 00:19:07,813 There’s some chance 737 00:19:07,813 --> 00:19:09,606 even money bet would say no. 738 00:19:09,857 --> 00:19:12,609 And you will have a life where 739 00:19:12,609 --> 00:19:14,236 anything you want to learn, 740 00:19:14,236 --> 00:19:16,488 you can learn very, very well. 741 00:19:16,488 --> 00:19:17,990 People will do different things 742 00:19:17,990 --> 00:19:18,699 with that. 743 00:19:18,699 --> 00:19:21,535 Some people will do cognitive offloading 744 00:19:21,535 --> 00:19:24,288 and actually become mentally lazier, 745 00:19:24,288 --> 00:19:27,416 and that will be a larger social problem. 746 00:19:27,416 --> 00:19:28,917 But the people who want to learn, 747 00:19:28,917 --> 00:19:30,794 it will just be a new paradise for them. 748 00:19:30,794 --> 00:19:32,171 I would say it is already. 749 00:19:32,171 --> 00:19:34,256 I consider myself one of those people. 750 00:19:34,256 --> 00:19:35,507 I’m happier. 751 00:19:35,507 --> 00:19:37,217 I use it to organize my travel, 752 00:19:37,217 --> 00:19:38,010 where I should eat, 753 00:19:38,010 --> 00:19:39,178 what book I should read. 754 00:19:39,178 --> 00:19:40,637 How to make sense of something in a book 755 00:19:40,637 --> 00:19:41,680 I don’t know. 756 00:19:41,680 --> 00:19:43,640 I’ll use it dozens of times every day, 757 00:19:43,640 --> 00:19:44,474 and it’s great. 758 00:19:44,641 --> 00:19:44,975 Yeah. 759 00:19:45,976 --> 00:19:46,393 Do you 760 00:19:46,393 --> 00:19:47,269 think that there’s a role 761 00:19:47,269 --> 00:19:49,521 for government or public policy, 762 00:19:49,521 --> 00:19:50,397 one way or the other, 763 00:19:50,397 --> 00:19:51,315 in smoothing 764 00:19:51,315 --> 00:19:52,774 the economic adjustment, 765 00:19:52,774 --> 00:19:55,027 let’s say, to widespread AI adoption? 766 00:19:55,569 --> 00:19:56,111 There will be. 767 00:19:56,111 --> 00:19:58,739 I’m not sure we know yet what it is. 768 00:19:58,739 --> 00:20:00,407 The immediate priority for me 769 00:20:00,407 --> 00:20:01,909 is just to get more energy 770 00:20:01,909 --> 00:20:03,493 supply out there 771 00:20:03,493 --> 00:20:05,954 so we can actually have a transition, 772 00:20:05,954 --> 00:20:08,540 since the transition now is so small. 773 00:20:08,540 --> 00:20:10,334 I’m a little wary of proposals 774 00:20:10,334 --> 00:20:12,628 to imagine 13 different problems 775 00:20:12,628 --> 00:20:14,213 in advance and solve them. 776 00:20:14,213 --> 00:20:15,339 I think we do better 777 00:20:15,339 --> 00:20:17,716 in the American system of government, 778 00:20:17,716 --> 00:20:20,052 often by waiting till problems come along 779 00:20:20,052 --> 00:20:21,386 when there’s high uncertainty 780 00:20:21,386 --> 00:20:22,971 and then addressing them piecemeal, 781 00:20:22,971 --> 00:20:24,223 whether it’s at federal, state, 782 00:20:24,223 --> 00:20:25,432 local level. 783 00:20:25,432 --> 00:20:26,808 So I’m skeptical 784 00:20:26,808 --> 00:20:27,809 if people who think they know 785 00:20:27,809 --> 00:20:29,478 how everything’s going to go 786 00:20:29,478 --> 00:20:30,771 and they have the magic formula 787 00:20:30,771 --> 00:20:31,855 for setting it right. 788 00:20:32,272 --> 00:20:33,565 I mean, I’m skeptical of that too, 789 00:20:33,565 --> 00:20:35,150 but that is our whole podcast, Tyler. 790 00:20:35,567 --> 00:20:36,193 I know. 791 00:20:36,193 --> 00:20:37,069 I understand. 792 00:20:37,152 --> 00:20:37,444 But the 793 00:20:37,444 --> 00:20:38,403 role of government, 794 00:20:38,403 --> 00:20:39,780 government laws are not 795 00:20:39,780 --> 00:20:41,323 easily repealed, right? 796 00:20:41,698 --> 00:20:43,033 If you pass something, 797 00:20:43,033 --> 00:20:44,868 you tend to stick with it. 798 00:20:44,868 --> 00:20:46,245 So I think we need to be 799 00:20:46,245 --> 00:20:47,454 a little cautious. 800 00:20:47,454 --> 00:20:48,497 Recognize things 801 00:20:48,497 --> 00:20:50,165 definitely will be needed. 802 00:20:50,165 --> 00:20:51,250 The first imperative 803 00:20:51,250 --> 00:20:52,793 will be better and different 804 00:20:52,793 --> 00:20:55,003 national security policies, 805 00:20:55,003 --> 00:20:56,797 which we’ve done some of that already. 806 00:20:56,797 --> 00:20:58,840 But those are tough, complicated issues 807 00:20:58,840 --> 00:21:00,884 no matter what your point of view. 808 00:21:00,884 --> 00:21:02,135 And then over time, well, 809 00:21:02,135 --> 00:21:03,887 a lot of our K-12 schools 810 00:21:03,887 --> 00:21:04,888 are governmental, 811 00:21:04,888 --> 00:21:06,265 so they should change. 812 00:21:06,473 --> 00:21:07,516 But it’s a big mistake 813 00:21:07,516 --> 00:21:08,725 to set out the formula. 814 00:21:08,725 --> 00:21:09,726 What you really want to do 815 00:21:09,726 --> 00:21:11,228 is create an environment 816 00:21:11,228 --> 00:21:12,020 where different states, 817 00:21:12,020 --> 00:21:13,563 counties, cities, experiment. 818 00:21:13,563 --> 00:21:15,524 And over time you figure out 819 00:21:15,524 --> 00:21:16,817 what is best and government 820 00:21:16,817 --> 00:21:17,943 schools should do it that 821 00:21:17,943 --> 00:21:19,361 way rather than all top down. 822 00:21:20,529 --> 00:21:21,738 It feels like that is advice 823 00:21:21,738 --> 00:21:23,198 that could apply to businesses too, 824 00:21:23,198 --> 00:21:24,741 as we try to advise leaders 825 00:21:24,741 --> 00:21:26,034 on how to prepare, 826 00:21:26,034 --> 00:21:27,119 even if this change 827 00:21:27,119 --> 00:21:28,370 comes from the outside in, 828 00:21:28,370 --> 00:21:30,122 even if it comes from those employees 829 00:21:30,122 --> 00:21:31,123 who are doing 830 00:21:31,123 --> 00:21:32,207 everything that you just said. 831 00:21:32,207 --> 00:21:35,711 You know, using an LLM 50 times a day, 832 00:21:36,586 --> 00:21:37,129 so much 833 00:21:37,129 --> 00:21:37,796 so that it becomes 834 00:21:37,796 --> 00:21:40,173 an ingrained thought partner. 835 00:21:40,924 --> 00:21:42,884 How should leaders 836 00:21:42,884 --> 00:21:45,887 think about adapting, adopting 837 00:21:46,221 --> 00:21:47,639 without getting locked in 838 00:21:47,639 --> 00:21:48,765 to something that, 839 00:21:48,765 --> 00:21:49,308 like you said, 840 00:21:49,308 --> 00:21:50,767 might change in six months? 841 00:21:51,393 --> 00:21:53,645 It’s going to change in six months. 842 00:21:53,979 --> 00:21:55,647 I think you have to teach your people 843 00:21:55,647 --> 00:21:57,024 how to stay current. 844 00:21:57,316 --> 00:21:59,067 Which is different from many other things 845 00:21:59,067 --> 00:22:00,402 you need to teach them. 846 00:22:00,402 --> 00:22:02,446 So all you had to teach people Excel. 847 00:22:02,446 --> 00:22:02,904 It’s true. 848 00:22:02,904 --> 00:22:04,823 Excel changes over time, 849 00:22:04,990 --> 00:22:07,242 but it’s been a fixed skill for a while. 850 00:22:07,242 --> 00:22:08,910 Using a smartphone, 851 00:22:08,910 --> 00:22:09,953 at least up until now, 852 00:22:09,953 --> 00:22:11,455 basically a fixed skill. 853 00:22:11,788 --> 00:22:13,415 We now have to teach people something 854 00:22:13,415 --> 00:22:14,624 that’s not a fixed skill. 855 00:22:15,417 --> 00:22:16,793 what about a college grad? 856 00:22:16,960 --> 00:22:18,003 Or, you know, like my son 857 00:22:18,003 --> 00:22:18,754 just started college, 858 00:22:18,754 --> 00:22:20,255 so they’re going to go— 859 00:22:20,255 --> 00:22:22,132 the ones who are leaving or just starting 860 00:22:22,132 --> 00:22:24,217 are in this very liminal space 861 00:22:24,217 --> 00:22:26,219 where the curriculum has not changed, 862 00:22:26,219 --> 00:22:27,012 but they will 863 00:22:27,012 --> 00:22:28,430 potentially come out into 864 00:22:28,430 --> 00:22:29,264 or are coming out 865 00:22:29,264 --> 00:22:31,266 into a changed environment. 866 00:22:31,266 --> 00:22:33,685 Is it enough that they have, 867 00:22:33,685 --> 00:22:35,771 you know, maximum neuroplasticity? 868 00:22:36,271 --> 00:22:37,064 They have to work 869 00:22:37,064 --> 00:22:38,857 much harder on their own, 870 00:22:38,857 --> 00:22:40,692 and there’s a kind of status adjustment 871 00:22:40,692 --> 00:22:42,652 many people will need to make. 872 00:22:42,652 --> 00:22:44,154 So the previous generation, 873 00:22:44,154 --> 00:22:44,738 there’s a sense 874 00:22:44,738 --> 00:22:45,405 if you’re smart, 875 00:22:45,405 --> 00:22:47,032 your parents have some money, 876 00:22:47,032 --> 00:22:48,742 you went to a good high school, 877 00:22:48,742 --> 00:22:49,868 that there’s a path for you. 878 00:22:49,868 --> 00:22:50,827 You can become a lawyer, 879 00:22:50,827 --> 00:22:51,453 go to finance, 880 00:22:51,453 --> 00:22:53,789 do consulting, work at McKinsey, 881 00:22:53,789 --> 00:22:54,831 and you will be 882 00:22:54,831 --> 00:22:56,291 like upper-upper-middle class 883 00:22:56,291 --> 00:22:57,125 or maybe better 884 00:22:57,125 --> 00:22:58,210 if you’re really talented. 885 00:22:59,169 --> 00:23:00,170 And I think that world 886 00:23:00,170 --> 00:23:01,963 is vanishing rapidly. 887 00:23:01,963 --> 00:23:03,548 The notion that these talented 888 00:23:03,548 --> 00:23:04,925 kids might have to forget 889 00:23:04,925 --> 00:23:06,343 about those paths 890 00:23:06,343 --> 00:23:07,052 and do something 891 00:23:07,052 --> 00:23:09,388 like work in the energy sector, 892 00:23:09,388 --> 00:23:11,640 which might even be for more money 893 00:23:11,640 --> 00:23:12,516 and move somewhere 894 00:23:12,516 --> 00:23:14,059 that’s not just 895 00:23:14,059 --> 00:23:16,269 New York or San Francisco, 896 00:23:16,269 --> 00:23:17,521 maybe, you know, move to Houston 897 00:23:17,521 --> 00:23:18,772 or Dallas 898 00:23:18,772 --> 00:23:21,108 or, you know, an energy center 899 00:23:21,108 --> 00:23:22,651 in Louisiana. 900 00:23:22,651 --> 00:23:24,361 It’s going to be a very different world. 901 00:23:24,361 --> 00:23:25,028 And I don’t think 902 00:23:25,028 --> 00:23:28,031 our expectations are ready for that. 903 00:23:28,240 --> 00:23:31,284 So the people who have invested 904 00:23:31,284 --> 00:23:32,285 in the current system 905 00:23:32,285 --> 00:23:34,121 and ways of succeeding, 906 00:23:34,121 --> 00:23:34,663 I think it will 907 00:23:34,663 --> 00:23:35,956 be very difficult for them. 908 00:23:35,956 --> 00:23:36,623 And a lot of these 909 00:23:36,623 --> 00:23:38,041 upper-upper-middle class 910 00:23:38,041 --> 00:23:39,209 parents and kids 911 00:23:39,209 --> 00:23:41,128 will maybe have a harder time 912 00:23:41,128 --> 00:23:41,711 than, say, 913 00:23:41,711 --> 00:23:43,755 maybe immigrants who came to this country 914 00:23:43,755 --> 00:23:45,006 when they’re eight years old, 915 00:23:45,006 --> 00:23:47,008 were not quite sure what to expect, 916 00:23:47,008 --> 00:23:48,176 and will, you know, 917 00:23:48,176 --> 00:23:49,719 grab at what the best options are, 918 00:23:49,719 --> 00:23:50,429 wherever they are. 919 00:23:51,847 --> 00:23:52,389 A lot of a lot of 920 00:23:52,389 --> 00:23:53,682 status adjustment across the board. 921 00:23:53,682 --> 00:23:54,558 It sounds like you’re saying, 922 00:23:54,558 --> 00:23:55,392 you know, from 923 00:23:55,392 --> 00:23:56,518 what your future is 924 00:23:56,685 --> 00:23:57,727 to the work that you do 925 00:23:57,727 --> 00:23:58,687 in an organization 926 00:23:58,687 --> 00:24:00,981 to how and who you manage. 927 00:24:01,356 --> 00:24:02,399 Like in economics, 928 00:24:02,399 --> 00:24:03,733 it used to be if you were very good 929 00:24:03,733 --> 00:24:05,652 at thinking like an economist, 930 00:24:05,652 --> 00:24:06,903 you knew that if you worked hard, 931 00:24:06,903 --> 00:24:07,737 you would do well. 932 00:24:07,946 --> 00:24:08,321 Uh-huh. 933 00:24:08,405 --> 00:24:09,990 I suspect moving forward, 934 00:24:10,323 --> 00:24:11,992 the AI can do that for you, 935 00:24:11,992 --> 00:24:13,368 and you have to be good at managing 936 00:24:13,368 --> 00:24:14,619 and prompting the AI 937 00:24:14,619 --> 00:24:16,413 and setting up research design 938 00:24:16,413 --> 00:24:18,248 in a way the AI can execute. 939 00:24:18,373 --> 00:24:19,791 That’s a very different skill. 940 00:24:19,791 --> 00:24:21,460 It’s not the end of the world, 941 00:24:21,460 --> 00:24:23,420 but when that change happens to you 942 00:24:23,420 --> 00:24:25,255 in the midst of your career plans 943 00:24:25,255 --> 00:24:26,548 it’s very disorienting. 944 00:24:26,798 --> 00:24:27,674 I just want to 945 00:24:27,674 --> 00:24:28,675 put a finer point on 946 00:24:28,675 --> 00:24:30,218 what I hear you saying a lot, 947 00:24:30,218 --> 00:24:30,635 which is 948 00:24:30,635 --> 00:24:32,220 and I think we know this, 949 00:24:32,220 --> 00:24:33,388 that energy, 950 00:24:33,388 --> 00:24:34,806 the energy conversation 951 00:24:34,806 --> 00:24:36,183 is a huge part of the 952 00:24:36,183 --> 00:24:37,559 AI future in a way that 953 00:24:37,559 --> 00:24:39,102 maybe we just have not been quite 954 00:24:39,352 --> 00:24:41,146 explicit enough about. 955 00:24:41,563 --> 00:24:43,064 I think we’ve not been very good 956 00:24:43,064 --> 00:24:43,940 at understanding 957 00:24:43,940 --> 00:24:45,817 it will need to expand a lot. 958 00:24:45,817 --> 00:24:47,694 We’ve made it very hard to build 959 00:24:47,694 --> 00:24:49,946 new energy infrastructure. 960 00:24:49,946 --> 00:24:50,780 I think that will again 961 00:24:50,780 --> 00:24:51,740 change over time, 962 00:24:51,740 --> 00:24:52,824 but probably a slow 963 00:24:52,824 --> 00:24:54,910 and somewhat painful process. 964 00:24:55,202 --> 00:24:56,119 But it’s also a place 965 00:24:56,119 --> 00:24:58,079 a lot of new jobs will be created there. 966 00:24:58,079 --> 00:24:58,663 A lot. 967 00:24:59,164 --> 00:25:00,790 Also biomedical testing. 968 00:25:00,790 --> 00:25:01,708 There’ll be all these new 969 00:25:01,708 --> 00:25:03,418 AI-generated ideas. 970 00:25:03,418 --> 00:25:05,128 Of course, most of them won’t work. 971 00:25:05,128 --> 00:25:06,922 Someone has to test them. 972 00:25:06,922 --> 00:25:08,924 Not just people working in research labs, 973 00:25:08,924 --> 00:25:10,258 but just actual people 974 00:25:10,258 --> 00:25:11,551 in the clinical trials. 975 00:25:11,551 --> 00:25:13,428 So that will be a much larger sector. 976 00:25:13,428 --> 00:25:14,638 That’s what will help us live 977 00:25:14,638 --> 00:25:16,556 to be 96 or 97. 978 00:25:16,556 --> 00:25:16,973 But again, 979 00:25:16,973 --> 00:25:18,099 you have to do the hard work 980 00:25:18,099 --> 00:25:19,017 in the meantime. 981 00:25:19,017 --> 00:25:20,810 And that will stretch on for decades. 982 00:25:20,810 --> 00:25:21,728 So people often 983 00:25:21,728 --> 00:25:23,813 ask me where the new jobs coming 984 00:25:23,813 --> 00:25:25,315 and part, I just don’t know. 985 00:25:25,315 --> 00:25:26,733 But those are two areas where, to me, 986 00:25:26,733 --> 00:25:27,901 it seems obvious 987 00:25:27,901 --> 00:25:29,277 there’ll be a lot of new jobs 988 00:25:29,277 --> 00:25:30,570 and they’ll be super productive. 989 00:25:30,904 --> 00:25:31,321 Right. 990 00:25:31,404 --> 00:25:32,155 And then 991 00:25:32,155 --> 00:25:32,614 I don’t want to 992 00:25:32,614 --> 00:25:33,490 put you on the spot further. 993 00:25:33,490 --> 00:25:34,866 You’re saying we’re early enough 994 00:25:34,866 --> 00:25:37,619 that we don’t even yet know what other 995 00:25:37,619 --> 00:25:38,703 I guess, cottage industries, 996 00:25:38,703 --> 00:25:39,955 so to speak, are going to be born 997 00:25:39,955 --> 00:25:41,164 as a result of this? 998 00:25:41,414 --> 00:25:41,873 That’s right. 999 00:25:41,873 --> 00:25:43,375 And which things 1000 00:25:43,375 --> 00:25:45,001 can the AI do, in essence, 1001 00:25:45,001 --> 00:25:46,002 all the work? 1002 00:25:46,169 --> 00:25:46,962 And in which things 1003 00:25:46,962 --> 00:25:47,837 will it still need 1004 00:25:47,837 --> 00:25:49,089 a lot of human guidance? 1005 00:25:49,089 --> 00:25:49,339 You know, 1006 00:25:49,339 --> 00:25:50,715 that’s going to be trial and error, 1007 00:25:50,715 --> 00:25:51,591 but that will determine 1008 00:25:51,591 --> 00:25:53,176 a lot where the new jobs go. 1009 00:25:53,176 --> 00:25:54,469 And in part, we just don’t know yet. 1010 00:25:55,095 --> 00:25:56,555 At the risk of 1011 00:25:56,555 --> 00:25:58,139 asking you to tell me, we don’t know yet 1012 00:25:59,766 --> 00:26:00,600 Which I will do. 1013 00:26:00,725 --> 00:26:01,977 I’m going in anyway. Exactly. 1014 00:26:01,977 --> 00:26:03,311 I trust that you will be honest with me 1015 00:26:03,311 --> 00:26:03,812 on this point. 1016 00:26:03,812 --> 00:26:04,563 But it’s how I tell you 1017 00:26:04,563 --> 00:26:05,063 we don’t know 1018 00:26:05,063 --> 00:26:05,522 that matters. 1019 00:26:05,605 --> 00:26:06,064 Exactly. 1020 00:26:06,064 --> 00:26:07,440 That’s the genius. 1021 00:26:07,774 --> 00:26:10,360 As we look forward three to five years, 1022 00:26:10,360 --> 00:26:11,945 I’m not going to ask you open ended 1023 00:26:11,945 --> 00:26:13,238 what you think things will look like. 1024 00:26:13,238 --> 00:26:13,989 That is unfair, 1025 00:26:13,989 --> 00:26:15,115 and nobody knows. 1026 00:26:15,365 --> 00:26:19,452 What would be the measuring sticks? 1027 00:26:20,161 --> 00:26:21,913 Like, what kinds of things 1028 00:26:21,913 --> 00:26:22,956 should we be measuring 1029 00:26:22,956 --> 00:26:24,124 to truly understand 1030 00:26:24,124 --> 00:26:26,251 the scale of economic impact 1031 00:26:26,251 --> 00:26:27,877 or change related to AI? 1032 00:26:28,336 --> 00:26:30,130 This may sound counterintuitive, 1033 00:26:30,130 --> 00:26:31,172 but I think 1034 00:26:31,172 --> 00:26:32,632 under a lot of scenarios, 1035 00:26:32,632 --> 00:26:35,260 the more unhappy people are, 1036 00:26:35,260 --> 00:26:36,094 actually the better 1037 00:26:36,094 --> 00:26:37,137 we’re doing, 1038 00:26:37,137 --> 00:26:38,972 because that means a lot of change. 1039 00:26:38,972 --> 00:26:39,931 Now there are scenarios 1040 00:26:39,931 --> 00:26:40,765 where people are unhappy 1041 00:26:40,765 --> 00:26:43,685 just because matters are terrible, okay. 1042 00:26:43,685 --> 00:26:44,728 But for the most part, 1043 00:26:44,728 --> 00:26:45,562 I think we’ll see this 1044 00:26:45,562 --> 00:26:46,563 negative correlation 1045 00:26:46,563 --> 00:26:49,232 between sentiment and actual progress, 1046 00:26:49,232 --> 00:26:50,984 because progress is disruptive 1047 00:26:50,984 --> 00:26:52,444 and there will be a lot of change. 1048 00:26:52,444 --> 00:26:53,653 That’s what I would look for. 1049 00:26:53,653 --> 00:26:55,113 So I’m not going to say, “I don’t know.” 1050 00:26:56,156 --> 00:26:56,573 All right. 1051 00:26:56,573 --> 00:26:57,824 Look for misery, everyone. 1052 00:26:57,824 --> 00:26:58,241 That will tell you 1053 00:26:58,241 --> 00:26:59,117 that it’s working. 1054 00:26:59,534 --> 00:27:00,493 But not quite misery. 1055 00:27:00,493 --> 00:27:01,870 It’s more like confusion 1056 00:27:01,870 --> 00:27:03,330 and disorientation 1057 00:27:03,330 --> 00:27:05,290 and some demoralization. 1058 00:27:05,332 --> 00:27:07,459 People, especially poorer people, 1059 00:27:07,459 --> 00:27:09,711 I mean, already have access to free, 1060 00:27:09,711 --> 00:27:12,172 very high-quality medical diagnosis. 1061 00:27:12,172 --> 00:27:13,465 If they need legal advice 1062 00:27:13,465 --> 00:27:14,841 to start a small business, 1063 00:27:14,841 --> 00:27:16,551 they don’t have to pay a lawyer. 1064 00:27:16,551 --> 00:27:17,594 Around the world, 1065 00:27:17,594 --> 00:27:18,637 in poorer countries, 1066 00:27:18,637 --> 00:27:20,930 this is starting to matter a great deal. 1067 00:27:20,930 --> 00:27:22,432 There’ll be so much more 1068 00:27:22,432 --> 00:27:24,017 happiness on net. 1069 00:27:24,017 --> 00:27:26,686 But if you just look at the complaining, 1070 00:27:26,686 --> 00:27:27,520 as we said before, 1071 00:27:27,520 --> 00:27:28,521 the complaining on Twitter 1072 00:27:28,521 --> 00:27:29,606 about all these different 1073 00:27:29,606 --> 00:27:32,150 AI-connected personas and companies, 1074 00:27:32,150 --> 00:27:33,026 the complaining 1075 00:27:33,026 --> 00:27:34,527 is going to go through the roof. 1076 00:27:34,527 --> 00:27:35,862 And that’s what I would measure. 1077 00:27:36,363 --> 00:27:37,447 That is outstanding. 1078 00:27:37,447 --> 00:27:39,240 Tyler Cowen is a professor of economics 1079 00:27:39,240 --> 00:27:40,450 at George Mason University, 1080 00:27:40,450 --> 00:27:43,161 author, and host of “Conversations with Tyler.” 1081 00:27:43,161 --> 00:27:44,245 And for more of his insights, 1082 00:27:44,245 --> 00:27:46,164 check out his blog “Marginal Revolution.” 1083 00:27:46,164 --> 00:27:47,540 Tyler, thank you so much. 1084 00:27:47,832 --> 00:27:48,625 Molly, thank you. 1085 00:27:49,292 --> 00:27:50,335 That was Tyler Cowan, 1086 00:27:50,335 --> 00:27:51,503 professor of economics 1087 00:27:51,503 --> 00:27:53,046 at George Mason University, 1088 00:27:53,046 --> 00:27:54,047 author and host of 1089 00:27:54,047 --> 00:27:55,715 “Conversations with Tyler.” 1090 00:27:55,715 --> 00:27:56,925 For more of his insights, 1091 00:27:56,925 --> 00:27:57,926 check out his blog, 1092 00:27:57,926 --> 00:27:59,219 “The Marginal Revolution.” 1093 00:27:59,219 --> 00:28:01,346 And for more conversations like this one, 1094 00:28:01,346 --> 00:28:02,222 follow our show, 1095 00:28:02,222 --> 00:28:03,431 and to find every episode 1096 00:28:03,431 --> 00:28:05,642 at microsoft.com/WorkLab. 1097 00:28:05,642 --> 00:28:06,976 I’m your host, Molly Wood. 1098 00:28:06,976 --> 00:28:07,894 Thanks for listening. 1099 00:28:09,145 --> 00:28:10,397 If you’ve got a question or comment, 1100 00:28:10,397 --> 00:28:11,356 drop us an email 1101 00:28:11,356 --> 00:28:13,191 at WorkLab@microsoft.com 1102 00:28:13,191 --> 00:28:13,733 and check out 1103 00:28:13,733 --> 00:28:15,652 Microsoft’s Work Trend Indexes 1104 00:28:15,652 --> 00:28:17,779 and the WorkLab digital publication, 1105 00:28:17,779 --> 00:28:19,280 where you’ll find all of our episodes 1106 00:28:19,280 --> 00:28:20,990 alongside thoughtful stories 1107 00:28:20,990 --> 00:28:21,574 that explore 1108 00:28:21,574 --> 00:28:23,243 how business leaders are thriving 1109 00:28:23,243 --> 00:28:24,869 in today’s AI era. 1110 00:28:25,078 --> 00:28:26,079 You can find all of that 1111 00:28:26,079 --> 00:28:28,164 at microsoft.com/WorkLab. 1112 00:28:28,164 --> 00:28:30,166 As for this podcast, rate us, review us, 1113 00:28:30,166 --> 00:28:31,710 and follow us wherever you listen. 1114 00:28:31,710 --> 00:28:33,712 It helps us a ton. 1115 00:28:33,712 --> 00:28:34,587 The WorkLab podcast 1116 00:28:34,587 --> 00:28:35,797 is a place for experts 1117 00:28:35,797 --> 00:28:37,841 to share their insights and opinions. 1118 00:28:37,841 --> 00:28:38,591 As students 1119 00:28:38,591 --> 00:28:39,467 of the future of work, 1120 00:28:39,467 --> 00:28:40,510 Microsoft values 1121 00:28:40,510 --> 00:28:42,721 inputs from a diverse set of voices. 1122 00:28:42,887 --> 00:28:44,180 That said, the opinions 1123 00:28:44,180 --> 00:28:46,099 and findings of our guests are their own, 1124 00:28:46,099 --> 00:28:47,809 and they may not necessarily reflect 1125 00:28:47,809 --> 00:28:50,311 Microsoft’s own research or positions. 1126 00:28:50,311 --> 00:28:51,855 WorkLab is produced by Microsoft 1127 00:28:51,855 --> 00:28:52,814 with Trifilm. 1128 00:28:52,897 --> 00:28:53,690 Sharon Kallander 1129 00:28:53,690 --> 00:28:54,733 and Matthew Duncan produce 1130 00:28:54,733 --> 00:28:56,484 the podcast with editorial support 1131 00:28:56,484 --> 00:28:57,902 from Jon Eilenberg. 1132 00:28:57,902 --> 00:28:59,779 Jessica Voelker is the WorkLab editor. 1133 00:29:00,196 --> 00:29:01,531 I’m your host, Molly Wood. 1134 00:29:01,531 --> 00:29:02,449 Thanks for listening.