1 00:00:02,180 --> 00:00:11,030 I realised that I'm just going to touch on one of the bullet points of roughly 20 something I governance challenges that Alan mentioned. 2 00:00:11,030 --> 00:00:15,950 So if you want to have your the relevance of your research being put in perspective, 3 00:00:15,950 --> 00:00:20,510 you go to the Future of Humanity Institute and you find out that you actually win just a very, 4 00:00:20,510 --> 00:00:26,210 very tiny small question related to all of the challenges that there are out there. 5 00:00:26,210 --> 00:00:36,330 And so the the topic that I'm going to focus on is what this artificial intelligence and automation and mean for the labour market. 6 00:00:36,330 --> 00:00:44,630 And for those of you who follow the debate and discussion, well, basically two very opposite views out there. 7 00:00:44,630 --> 00:00:49,520 One is that there's going to be this employment apocalypse. 8 00:00:49,520 --> 00:00:54,590 Artificial intelligence is going to take all the jobs. I'm commonly associated with that one. 9 00:00:54,590 --> 00:01:01,130 And there's the other perspective, which is that if we only look to history and we should feel quite reassured 10 00:01:01,130 --> 00:01:07,970 because automation has been progressing for centuries and perhaps even millennia. 11 00:01:07,970 --> 00:01:15,020 And what I tried to do in my recent book is actually to go back and look to history 12 00:01:15,020 --> 00:01:21,140 and what has happened to people and their lives as the jobs where and displaced. 13 00:01:21,140 --> 00:01:29,690 And the question I try to ask is, should we feel that reassured if the future of automation mirrors the past? 14 00:01:29,690 --> 00:01:37,460 Now I'm sort of covering roughly 10000 years of history in the book, and I'm not going to cover all of that in 15 minutes, 15 00:01:37,460 --> 00:01:44,640 but I intend to at least give you sort of a glimpse of the things that are covered in the book. 16 00:01:44,640 --> 00:01:50,480 And so and the first story in the book has to do with the fates of the Lamplighter 17 00:01:50,480 --> 00:01:56,240 and several hundred years ago and the streets of Oxford weren't around this time. 18 00:01:56,240 --> 00:02:05,480 The lit by lamp like this, walking around the streets with air torches and letters and and two lamp artists. 19 00:02:05,480 --> 00:02:11,090 The arrival of electric lights was an enormously disruptive event. 20 00:02:11,090 --> 00:02:18,680 And as electricity that was increasingly regulated from substations, the jobs were made redundant. 21 00:02:18,680 --> 00:02:27,590 And for lamplight, this who essentially supported themselves and their families and and on these jobs, 22 00:02:27,590 --> 00:02:32,120 they were naturally not that happy with what was happening. 23 00:02:32,120 --> 00:02:41,870 And in Brussels in 1987, Lamplighter took the streets in fear of losing their jobs and smashed the electric streetlights. 24 00:02:41,870 --> 00:02:48,410 And this situation escalated, with the police being sent out the loud protests and an operating police headquarters, 25 00:02:48,410 --> 00:02:53,390 and the army had to be sent in to resolve the situation. 26 00:02:53,390 --> 00:03:01,730 Well, there's a tendency to ridicule people like the lamp of being backward and Luddites. 27 00:03:01,730 --> 00:03:09,920 And when we do this, we tend to focus very much on the very long run of technological and economic progress. 28 00:03:09,920 --> 00:03:16,100 So up until roughly 18:00, growth across the world was fairly stagnant. 29 00:03:16,100 --> 00:03:25,460 And that is not to say that nothing happened. There was a lot of innovation and progress going on, but per capita incomes didn't rise significantly, 30 00:03:25,460 --> 00:03:32,360 and incomes per capita did only really take off when the first industrial revolution with the first 31 00:03:32,360 --> 00:03:37,610 machine age with the mechanised factory that allowed us to produce with more with fewer people. 32 00:03:37,610 --> 00:03:47,660 And as a result of mechanisation over the past 200 years, the average person in Britain is now roughly three times better off than they were in 18:00. 33 00:03:47,660 --> 00:03:59,930 Adjusted for inflation. And if we only look at incomes that, if anything, understates the transformation that has taken place because needless to say, 34 00:03:59,930 --> 00:04:08,780 the consumer baskets the people have access to today is very different from the consumer basket that people have access to in and out of it. 35 00:04:08,780 --> 00:04:16,160 So most people then could only dream about all of these goods and services that are now part of everyday life for us. 36 00:04:16,160 --> 00:04:20,360 And a lot of these things are not even accounts for the GDP statistics. 37 00:04:20,360 --> 00:04:24,590 So consider, for example, what you would pay for anaesthesia. 38 00:04:24,590 --> 00:04:29,150 If you undergo heart surgery, the mouth would be almost infinite, right? 39 00:04:29,150 --> 00:04:33,980 And its contribution to GDP is basically zero. 40 00:04:33,980 --> 00:04:38,090 And if that is not evidence of a lot of progress, 41 00:04:38,090 --> 00:04:42,830 consider the fact that producing those higher incomes and producing all of 42 00:04:42,830 --> 00:04:47,840 these additional goods and services has become a lot more comfortable as well. 43 00:04:47,840 --> 00:04:53,480 And during the first industrial revolution, a significant share of the workforce was in coal mines. 44 00:04:53,480 --> 00:04:56,960 Caverns and explosions were part of everyday working life long. 45 00:04:56,960 --> 00:05:02,250 The sea is often part of the work package. Today, most of us work in air conditioned offices. 46 00:05:02,250 --> 00:05:06,810 And if we look at sort of the sector transformation that, if anything, 47 00:05:06,810 --> 00:05:12,750 understates the progress that has taken place because a lot of individual jobs have been transformed as well. 48 00:05:12,750 --> 00:05:20,070 And so back in nineteen hundred at farm labour over the fields for nothing more than animal power today, 49 00:05:20,070 --> 00:05:26,790 a farm labour in Britain was set in his or her tractor and listen to the music of his choice. 50 00:05:26,790 --> 00:05:36,900 So working life has gotten a lot better. The question is what's happened to all those people that went through this transformation? 51 00:05:36,900 --> 00:05:47,580 And if we go back to the industrial revolution and look what people said about it back then, I think we can find some instructive answers. 52 00:05:47,580 --> 00:05:54,870 And so in eighteen forty eight, Benjamin Disraeli, before he became prime minister of Britain, 53 00:05:54,870 --> 00:06:00,060 published a novel in which one character remarks that I see cities people with machines. 54 00:06:00,060 --> 00:06:04,530 Suddenly, Manchester must be the most wonderful place of modern times. 55 00:06:04,530 --> 00:06:11,730 And the very same year, Frederick Engels published his book on the conditions of the working classes, 56 00:06:11,730 --> 00:06:19,020 which was written during precisely a stay in Manchester, and an analyst had a very different take on the matter. 57 00:06:19,020 --> 00:06:25,110 He argued that machine rarely set to downgrade people. It puts them in the repetitive motions of the machine, 58 00:06:25,110 --> 00:06:34,980 which he deemed to be a natural and only serves to put pressure on workers wages and potentially even displace them from their jobs. 59 00:06:34,980 --> 00:06:40,320 Now we all know that he was not quite on target about the future, 60 00:06:40,320 --> 00:06:48,720 but he was actually fairly on target about the period he lived through because for roughly seven decades, even as the British economy took off. 61 00:06:48,720 --> 00:06:54,030 Wages are stagnant and probably even falling at the lower end of the income distribution. 62 00:06:54,030 --> 00:07:02,820 And the wage data for this period is not great. But if you look at other sources of data like consumption or biological and indicators of well-being, 63 00:07:02,820 --> 00:07:10,530 such as heights, we find that the courts born in 1850 were actually shorter than the course, all in 1750, 64 00:07:10,530 --> 00:07:16,440 and part of the reason for that is that people's nutrition was adversely impacted as they were 65 00:07:16,440 --> 00:07:24,690 displaced from domestic and industry and passing to economists and economic historians has always been. 66 00:07:24,690 --> 00:07:33,510 Why would people voluntarily have agreed to participate in the industrialisation process if it reduced their utility? 67 00:07:33,510 --> 00:07:36,570 Well, the simple answer to that is that they did not. 68 00:07:36,570 --> 00:07:45,570 They rioted against the mechanise factory on several occasions, their petition to parliament to block the introduction and of machinery. 69 00:07:45,570 --> 00:07:53,880 And how did the British government respond? Well, on several occasions, actually, by sending our troops against the riots and the army, for example, 70 00:07:53,880 --> 00:08:03,640 that was sent out as a Luddites was larger than the army that Wellington took against Napoleon and the Peninsular War in 18 08, 71 00:08:03,640 --> 00:08:13,530 and the love that riots that will tend to focus on, well, part of a long wave of riots that swept across Britain, continental Europe, India and China. 72 00:08:13,530 --> 00:08:21,540 Resistance to mechanisation has actually been more of the historical norm, rather than the exception. 73 00:08:21,540 --> 00:08:27,930 Now the point is obviously not that, you know, we are about to live through all of this history again. 74 00:08:27,930 --> 00:08:32,820 But I think it's at least put some of the current concerns in perspective. 75 00:08:32,820 --> 00:08:39,540 And what we do see today is that since roughly the nineteen eighties with the computer revolution, 76 00:08:39,540 --> 00:08:46,920 inequality in Britain and the United States has been approaching levels not seen since the first industrial revolution. 77 00:08:46,920 --> 00:08:52,920 And clearly, there are many variables that have shaped patterns of inequality. 78 00:08:52,920 --> 00:08:58,050 But technology is certainly one of the key factors. 79 00:08:58,050 --> 00:09:06,100 And and the technology today could not be more different than the steam powered machines of the first industrial revolution. 80 00:09:06,100 --> 00:09:11,370 But what they do have in common with computers is that they are replacing in a 81 00:09:11,370 --> 00:09:18,420 variety of tasks the way that spinning machines replaced artisan craftsmen. 82 00:09:18,420 --> 00:09:29,400 Computers are now replacing a variety of jobs and in the labour market, and it is certainly true that computers have been with us for a long time. 83 00:09:29,400 --> 00:09:39,690 The first electronic computer was developed at the University of Pennsylvania and in 1947, but it consisted of 18000 vacuum tubes and 30 tons. 84 00:09:39,690 --> 00:09:43,740 And as a result of that, it didn't have much impact on the labour market. 85 00:09:43,740 --> 00:09:50,160 And this has been true with all technologies. It's taken a long time for the technology to become sufficiently good and 86 00:09:50,160 --> 00:09:56,970 sufficiently cost effective to have a real impact and on jobs and people's lives. 87 00:09:56,970 --> 00:10:01,810 And for some of us, computerisation has been great for me as an academic. 88 00:10:01,810 --> 00:10:09,850 I could do more statistical analysis, I can write more papers and I can export my ideas and thoughts to the world. 89 00:10:09,850 --> 00:10:12,490 And but for a lot of people, 90 00:10:12,490 --> 00:10:23,080 especially those with lower levels of skills and especially men who used to work on the assembly lines in factories, it hasn't been great. 91 00:10:23,080 --> 00:10:29,530 And if you look at data from the United States, what you see is not just that inequality is rising, 92 00:10:29,530 --> 00:10:39,130 but the wages of prime aged men with no more than a high school degree has actually been falling in real terms since the 1980s. 93 00:10:39,130 --> 00:10:44,230 And as I mentioned earlier, the income distribution is shaped by a lot of factors. 94 00:10:44,230 --> 00:10:51,790 But if you want to explain why particularly prime aged men are lower levels of education have seen their wages fall. 95 00:10:51,790 --> 00:10:59,800 Globalisation and automation are the prime factors, and of course, they're very much interrelated without computers. 96 00:10:59,800 --> 00:11:08,650 It would be completely unfeasible for businesses to relocate or restructure supply chains in ways that they can take advantage of cheap labour. 97 00:11:08,650 --> 00:11:16,480 And in countries like China, and we have already seen a backlash against globalisation. 98 00:11:16,480 --> 00:11:21,310 But automation has had very similar effects on the labour market. 99 00:11:21,310 --> 00:11:26,290 And if you want to understand why President Trump won three key swing states that 100 00:11:26,290 --> 00:11:31,270 they've been won with it by the Democratic candidate every election since 1992. 101 00:11:31,270 --> 00:11:39,820 And automation is one of the key reasons Michigan, Wisconsin, Pennsylvania old Rust Belt states, 102 00:11:39,820 --> 00:11:45,580 which have been both heavily hit by automation and globalisation. 103 00:11:45,580 --> 00:11:49,570 You can debate the relative merits of these two variables, 104 00:11:49,570 --> 00:11:55,930 but they are both driven by technology, and they all both reflect what has already happened. 105 00:11:55,930 --> 00:11:59,470 Looking forward, the rise of China has taken place. 106 00:11:59,470 --> 00:12:02,350 It's not going to happen one more time. 107 00:12:02,350 --> 00:12:13,720 But looking forward, the potential scope of automation is getting much bigger because all of this relates to sort of the RULE-BASED era of computing, 108 00:12:13,720 --> 00:12:20,830 with the programmer specifying what the technology should do at every given contingency, 109 00:12:20,830 --> 00:12:29,890 with various machine learning techniques becoming more pervasive but gradually sort of going beyond the automation of 110 00:12:29,890 --> 00:12:39,160 routine rule based activities into a variety of things like machine translation like diagnostics like document review, 111 00:12:39,160 --> 00:12:45,130 like even potentially driving cars. And beyond artificial intelligence. 112 00:12:45,130 --> 00:12:52,710 I think one reason that a lot of people underestimate the potential scope of automation is that it's not. 113 00:12:52,710 --> 00:12:59,200 The technology doesn't need to be capable of doing what you do in your job in order to replace you. 114 00:12:59,200 --> 00:13:04,460 We didn't automate away the jobs of land places by building robots capable of climbing lampposts. 115 00:13:04,460 --> 00:13:11,290 What didn't automate the way the jobs of non-racist by building robots that would walk out of the house shot down the words 116 00:13:11,290 --> 00:13:17,230 carry buckets of water and wood into the home and heat it on the stove and then perform the motions of hand-washing. 117 00:13:17,230 --> 00:13:20,770 We did that by inventing an electric washing machine. 118 00:13:20,770 --> 00:13:28,150 Now I've spent about eight years of my life discussing the 47 percent and the future of automation, 119 00:13:28,150 --> 00:13:36,250 so I'm only going to mention the fact that we think that a lot of occupations and industries and are in fact affected by the expanding 120 00:13:36,250 --> 00:13:46,660 scope of automation in particularly retail and office support tasks and tasks related to transportation and material moving. 121 00:13:46,660 --> 00:13:53,200 And when we published this study 80 years ago now roughly. 122 00:13:53,200 --> 00:13:59,740 And we also published a very detailed list, about 700 to occupations and there are lots of exposure. 123 00:13:59,740 --> 00:14:08,230 So automation, so everybody can look at sort of what those predictions are and can compare retrospectively how things are going. 124 00:14:08,230 --> 00:14:13,600 And one, occupations that we were constantly tasteful was that we found that fashion models 125 00:14:13,600 --> 00:14:18,350 and are exposed to automation and the models on this picture actually don't exist. 126 00:14:18,350 --> 00:14:24,700 They've been created through generative adversarial networks and they actually have their own Instagram accounts. 127 00:14:24,700 --> 00:14:35,320 Now, needless to say, the future of the labour market is not just going to be determined by the first order effects on automation, 128 00:14:35,320 --> 00:14:40,030 they're going to be a lot of second and third order effects, the much more harder to predict. 129 00:14:40,030 --> 00:14:46,160 But we know that early success and historic right. So when factories electrified early on, for example, 130 00:14:46,160 --> 00:14:52,930 all that's factory on the street was replacing the steam engine with an electric motor as the central power source of the factory. 131 00:14:52,930 --> 00:14:56,020 Well, this shifts and counter chests remain intact. 132 00:14:56,020 --> 00:15:01,590 And it took engineers a little while to figure out that while you can actually equip every single machine with. 133 00:15:01,590 --> 00:15:09,600 His own electric motor, and then you can sequence those machines, according to national flow production, which gave rise to mass production, 134 00:15:09,600 --> 00:15:16,920 allowed Henry Ford to be produced a Model T as a at a sufficiently low price for it to become the people's vehicle. 135 00:15:16,920 --> 00:15:19,800 And those techniques then spread from industry to industry. 136 00:15:19,800 --> 00:15:26,970 That's an enormous expansion of manufacturing as prices come down and the demand for manufacturing goods went up. 137 00:15:26,970 --> 00:15:32,880 And if you look at the first automobiles, there basically looks like a horse carriage rides. 138 00:15:32,880 --> 00:15:38,940 So what we did was essentially replacing the horse with an internal combustion engine and to drive it. 139 00:15:38,940 --> 00:15:43,280 And it took a while for people to figure out that we needed a complementary infrastructure. 140 00:15:43,280 --> 00:15:51,960 We need traffic laws and have raised a lot of jobs in retail, in and around commerce and so on and so forth. 141 00:15:51,960 --> 00:15:57,970 All of these were secondary effects that nobody could imagine at the time, but it really reshaped the economy. 142 00:15:57,970 --> 00:16:07,590 And but all of this in the end of the day depended on the exception and of this new technologies. 143 00:16:07,590 --> 00:16:12,030 And I do think that Leon TIFF was onto something when he suggested that if a 144 00:16:12,030 --> 00:16:15,390 horse is going to join the Democratic Party and voted what happened on the farm, 145 00:16:15,390 --> 00:16:22,650 it might have turned out differently. They could have used the political clout to bring the spread of the factory to a halt. 146 00:16:22,650 --> 00:16:31,110 And this is essentially what a lot of us tried to do, and a lot of people did quite so successfully for centuries before, 147 00:16:31,110 --> 00:16:37,830 because usually governments sided with angry workers rioting against the mechanised factory. 148 00:16:37,830 --> 00:16:43,890 If you want to understand why Britain was first industrialised rather than France, this is actually one of the prime reasons. 149 00:16:43,890 --> 00:16:50,160 And because machinery resistance in France occurred very much also during the really revolutionary era, 150 00:16:50,160 --> 00:16:58,560 but governments didn't have the political clout to squash the machine riots in the same way that the British government did. 151 00:16:58,560 --> 00:17:05,340 And you see, also in China, machinery riots were actually quite successful up until the turn of the 20th century. 152 00:17:05,340 --> 00:17:11,190 So this sort of historical patent has actually been one of resistance to these technologies. 153 00:17:11,190 --> 00:17:13,170 Now do we see this happening again? 154 00:17:13,170 --> 00:17:20,940 Well, we can at least highlight some examples, so we didn't actually suggest that 47 percent are planned to be automated. 155 00:17:20,940 --> 00:17:31,590 It's not a conspiracy, but these are nonetheless and people striking against the introduction of autonomous cargo trucks in Los Angeles. 156 00:17:31,590 --> 00:17:39,060 These are truck drivers in the state of Missouri demanding legislation to block the introduction of autonomous trucks. 157 00:17:39,060 --> 00:17:45,540 And more broadly, a recent Pew Research survey suggests that the majority of Americans now think that there 158 00:17:45,540 --> 00:17:51,420 should be limits to the number of machines that businesses should be allowed to implement. 159 00:17:51,420 --> 00:17:57,420 And so do I think that, you know, we're going to see a lot of resistance to mechanisation in the future. 160 00:17:57,420 --> 00:18:01,470 The simple answer is, I don't know. But it has been the historical norm. 161 00:18:01,470 --> 00:18:09,540 And if people don't see the benefits of technology in the short run, the show may have some incentives to resist it. 162 00:18:09,540 --> 00:18:14,250 And we should also remember that when economists speak about the short run and not very 163 00:18:14,250 --> 00:18:18,390 specific about what the short term is and this kind of matters whether it's 20 hours, 164 00:18:18,390 --> 00:18:25,200 20 days or 20 years in the context of the first industrial revolution, it was seven decades since the 1980s. 165 00:18:25,200 --> 00:18:31,560 Wages for certain groups in the labour markets falling consistently. And it's not my sort of health this state of affairs. 166 00:18:31,560 --> 00:18:41,430 So I think we need to sort of think about what policies we need to help people and in this short run, and I do touch upon them in the book as well. 167 00:18:41,430 --> 00:18:48,042 I think I'm probably running out of time, so thank you very much.