Automation, management, and the future of work
01:27:31 Share

Welcome to the LSE Events podcast by the London School of Economics and Political Science. Get ready to hear from some of the most influential international figures in the social sciences.

Good evening everybody. I am Professor Sarah Ashwin and I welcome you to tonight's discussion hosted by the LSE's Department of Management where I serve as Head of Department. I'm really pleased to welcome both our in-person and online audience to what I'm hoping will be a really exciting discussion tonight on automation management and the future of work.

The impact of new technology on employment, management, societies and politics is a crucial contemporary issue.

Those of you who know LSE well may recall that our President and Vice-Chancellor, Professor Larry Kramer, recently spoke on this topic at his inaugural lecture where he identified the emergence of new technologies as one of the five key challenges for social sciences.

We are therefore very lucky to hear from Professor Eric Hurst this evening, an economist whose work lies at the intersection of macroeconomics, labour economics and urban economics.

Eric visits us tonight from the University of Chicago Booth School of Business. He also serves as a member of the Economics Fluctuations Group, Aging Group and Public Economics Group at the National Bureau of Economic Research in the US.

Eric is interested in the intersection of new technologies and labour displacement and tonight he will explore how technology will impact demand for labour or to put it more colloquially, whether he'll think about whether robots are coming for your jobs. It's immediately obvious that this topic has huge social and political implications.

I'm therefore delighted that to help explore these issues, Eric is joined on stage by two scholars. First of all, Chrysanthi Avgaru, Professor of Information Systems at LSE's Department of Management. Chrysanthi's research focuses on information technology, innovation and organisational change. And Joanne Yachtman,

Drummond Professor of Political Economy at the University of Oxford, whose recent research has focused on AI, labour economics, as well as political questions such as protest patterns and the relationship between AI development and autocracy. So, as you can see, our discussants are going to bring different perspectives to a topic which is exciting, but also potentially a little bit scary in its implications. So,

Before I introduce or offer the floor to our speakers, a few words on housekeeping. As usual, there will be a chance for you to put your questions to our speakers. We will try to ensure that a range of questions from both our online and our audience here in the theatre are taken. So we'll go to both groups.

Those of you here in the theatre, please raise your hand and someone will bring you a microphone. Please don't speak until the person with the microphone arrives because otherwise our online audience won't be able to hear you. When called upon, please give us your name and affiliation before posing one short question. That is, I know you're all very clever, but no mini lectures please, just questions.

Those of you joining us online can submit your questions through the Q&A feature at the bottom of your screen. Please also let us know your name and affiliation. We're particularly keen to hear from our students and our alumni. Finally, for social media users following the conversation, the hashtag for today's event is #LSEevents, all one word.

The event is being recorded and subject to no technical difficulties because sometimes the robots do have problems, luckily for us. As long as there are no such problems, this will be made available as a podcast. So now, will you please join me in welcoming Professor Eric Hurst. APPLAUSE

Thank you so much. It's so good to be here. I love coming to LSE. It is one of my favorite places to visit academically. So what I'm showing here to start off is just a

picture of the employment rates for men between the ages of 25 and 55 in the United States over time. And so the first thing that you should be seeing is that this trend in employment rates has been falling relative to kind of where it was in the 1970s, 1980s, early 1990s. It used to be about 90%

of men in 25 to 55 year old range tended to work. That current number in the United States is about 86% right now. And if you put that into people, there's about 2 million less men working now than there would have been in 20 to 25 years earlier.

This next part I just want to focus on. I give these talks often. And this is, I would say, the one that gets me most depressed when I see it. So what I'm looking at here, there's two lines. The top line is men with less than a bachelor's degree. The bottom line is men with a bachelor's degree or more in that same age range, 25 to 54. And what I'm measuring is the share that reports

working zero weeks per year in the prior year. And so historically, for men with less than a bachelor's degree, that number used to be about 8%, going back to the mid-1980s. That number today is somewhere around 14% or so.

So roughly, one on seven of men in this age range with less than a bachelor's degree is sitting idle from the labor force for a whole year. You see a trend in those with a bachelor's degree or more, but much more muted in the level overall is lower.

Again, I showed you pictures now for men. The women pictures is confounded by long run trends, but I do want you to notice that that trend in employment rates seems to have stagnated starting in the early 2000s. So if we go back to the history of time in the United States, female labor supply has been increasing. The percentage of women who report working has been increasing.

That number was about somewhere around 75% in the early 2000s. That number has roughly stayed that way since. So that trend has roughly stopped.

And so what I want us to be thinking about, or at least what I'm going to talk about in my portion of our time together tonight, is try to think about factors that might be behind some of these trends. And I'm going to focus in particular on technology. We're going to go back and forth between using the words technology and robots and automation and AI, and they're all similar sides of the same coin. And when I get towards the end of today's talk, I want to try to give us a framework to think about these trends.

technological changes broadly. But I'm going to start by giving us a little bit of discussion of automation in one particular sector in the United States and I'm going to show you some data from the UK as well, the manufacturing sector. And so has there been changes in the manufacturing sector that could be related to these changes in employment that we've seen so far? I am then going to move and talk a little bit about

some prior time periods. Like we had automation in the past in some sectors. Agriculture. And what happened during that time period. And then I'm going to try to end with a little bit of framework-ish stuff, questions to be thinking about, about the labor market effects of automation. And then we'll move to our discussion collectively.

So let me show you some patterns on the, this is the level of manufacturing employment in the United States going back to the late 1970s again. And so the numbers you can see, it might be hard to see, but there's about historically in the 1970s, about 18 million men working, this is not men, all groups working in the manufacturing sector in the United States in the 1970s.

And so that's been declining over time. And so we've seen like about 2 million manufacturing jobs disappear in the 1980s in the United States. And there's movies written about the decline of Detroit. That's that period where manufacturing employment was going into some manufacturing towns where Halloween happened. I want to contrast that, what happened to the level of manufacturing employment starting in the early 2000s.

We lost about 6 million jobs in manufacturing in the United States during the 2000 to 2010 period. Many of those jobs disappeared well before the start of the Great Recession. And since, there's been very little rebound. So it's not like those manufacturing jobs have come back.

And so this is the same kind of patterns, I'm not going to go through them in as much detail, for also Germany and the UK. This is not a US phenomenon. Manufacturing, and this is the share of employment in manufacturing, has been declining in all countries. And you can see it's actually steeper, the decline, the blue line in the United Kingdom than it is even in Germany or the United States. So manufacturing jobs are getting less and less prevalent in the economy.

And so there's stories that people have talked about about why those manufacturing declines have been occurring. And there's two stories, and I believe those two stories are related.

Story one is exposure to trade, that there has been a country, we'll call it China, that's been in the background that has now specialized in manufacturing. It is a large country and has been exporting lots of manufacturing goods around the world, and that means the production of manufacturing in countries like the United States or the UK might be less than it would before because we have a competitor from abroad.

Story two is automation is occurring. And automation means for a given amount of manufacturing production, we could do it with less workers and more machines. And those two stories are intrinsically linked. There is research out there showing that the sectors that were most exposed to trade with China were also the sectors that automated the most.

Now, how do we measure the automation? Well, I'm going to give you one moment of that measurement of the automation. I'll show you one statistic that I'm going to look at. We call this, again in geek economics talk, the labor share of income in the manufacturing sector. That means for a dollar earned by manufacturers, how much of that goes to workers and how much of it goes to machines? And so this is the manufacturing share of income in the United States.

in the manufacturing sector, again going back to the mid-1980s in this picture through today. And it used to be that about one-third of all income in the manufacturing sector went to workers. That number today is somewhere around one quarter. About somewhere a seven, eight percentage point decline in the share that's going. That means less going to workers, the firms are producing stuff, but they're doing so more with the machines and less with the workers.

And I should have mentioned earlier when I had the employment numbers up there, the composition of employment in the manufacturing sector is also shifting dramatically. We're using less production workers and more engineers. So the share of employment of production workers is falling even more sharply than it is in total employment because the composition towards the R&D and engineers and the manufacturing sector is going up. And the patterns are very similar in the UK and the US in these changes in the labor share.

So automation is somehow going on in this sector. So now you might ask yourself, how can we link this automation to labor market outcomes like employment rates or earnings in the economy?

And so the way we do this, and there's some of us in the room who've kind of done this in the past, we can exploit regional variation within a country to see if there's a potential link between manufacturing employment changes and total employment changes in the economy. So let me show you a picture.

And this is the United States. There's 700 little grids in this picture. Those are called commuting zones in the United States. A commuting zone is a geographically defined area where the people who live in that area predominantly work in that area. I'm from Chicago. There's a Chicago commuting zone. It's this little orange one right there. Most people in that area work in that area. And there's 700 of those in the United States. The United States is big.

The deepness of the red is how intensive manufacturing employment was in that commuting zone in 2000, before the rise of China. And you can see the places where manufacturing is concentrated in the United States isn't equally dispersed across the United States. It is in places like-- you might have heard of these places recently-- Michigan, Pennsylvania, Wisconsin. Georgia has some as well.

And so these are places that were heavily, heavily manufacturing in the early 2000s. And so when manufacturing declines, it is going to be disproportionately concentrated in the places that did manufacturing. And so then we could use cross-region variation to kind of get a sense of what happened to employment rates overall in the places where manufacturing declined.

like Michigan, versus places where manufacturing was not as predominant, like in Florida. They produce like Disney World and vacations and things of that nature. And so then we can kind of get a sense about, I'm just going to show you a correlation now, it's not causation, but I'm going to show you a correlation between each one of these circles is a commuting zone. The size of the circle is how big that commuting zone is. So the bigger circles are New York and LA and Chicago. The smaller circles are going to be places in Duluth,

Minnesota, so smaller places. On the x-axis is how much manufacturing declined in that region between 2000 and 2015 as a share of employment. The y-axis is how much the total employment rate fell in those regions

during that same time period. And what you can see is a sharp downward sloping line. The places that were more exposed to manufacturing declines are the places in the United States that had the worst labor market outcomes, where employment rates were falling, wages were falling in those same places as well.

So you can see the Michigans and the Wisconsin's and the Pennsylvania's. Those are the places that have been suffering as manufacturing has been declining. This picture is-- I'm showing you this for men, all men. It's the same for women. And the line is much deeper if I look at men without a bachelor's degree relative to men with a bachelor's degree. So it's concentrated in parts of the distribution where accumulated levels of schooling are less.

There are similar, it's not my own, this is my own work, but I looked online just when I was putting this lecture together, there are similar patterns for Britain as well. The graph is inverted, so manufacturing decline is on the y-axis here. So in the West Midlands, there was a 13% touch point decline in manufacturing employment from the years mid-1990s through today. And on the y-axis is how much real earnings per capita has fallen in those places.

And so a number like 90 means a 10% decline relative to baseline. And so you can see the places that were in the UK where manufacturing was declining the most are also having adverse labor market outcomes. Okay, so there seems to be something about this technological shift, automation going on, where at least in the period of the last two decades in the manufacturing sector that is associated with declining employment rates.

Is that a normal phenomenon? Let me kind of show you some quotes. This is a good one. "We are being afflicted with a new disease of which some readers may not have heard the name, but of which they will hear a great deal in the years to come. Namely, technological unemployment." John Maynard Keynes, 1930. "Labor will become less and less important. More and more workers will be displaced by machines. I don't see that new industries could ever employ. Everybody wants a job.

Wassily Leontief, 1952. That guy won a Nobel Prize as well. So these discussions of technology and labor market have been occurring for centuries. Now what were these guys talking about 70, 100 years ago? They were talking about a robot. Well, they didn't call it a robot. It was a tractor. And that tractor came to the agriculture sector and displaced a whole bunch of workers. So in the United States, if we go back to 1910, one third of all men in the United States worked in agriculture.

That number today, 3%. Machines came along, automation came along, and that automation was displacing workers. But yet, we weren't talking about 1970, 1980, 1990, the economy unraveling. Those workers in those sectors migrated somehow to other sectors that were recurring jobs.

And so in my last part, before we move to our panel where we can kind of talk a little bit more deeply about some of these issues, is I want to just give us a little things that I think about, or at least when I teach my students or write in my papers, about how economies adjust to technological change in the labor market. So there's three questions that we need to ask the answer about the labor market effects of technology and in technology. AI, robots, machines in the manufacturing sector.

Attractor. Question one. Is the technology, I'm going to use geeky languages for a second, a complement or a substitute with the workers in the labor market? What do I mean by that? Does the technology make firms want to hire us more because we're more productive? Technology comes along. Again, back in the day, I was writing papers and, you know,

computers were around but they weren't that prevalent and I would have to you know write and you'd send your co-author something by mail and they'd send you stuff back and then you'd kind of go through computers made us more productive we could actually write on the same files at the same time so in some sense technology may be more productive that is a compliment with my productivity

And other times, the technology might be a substitute. The robot comes in and does exactly the same job, so the workers don't need me, or the firms don't need me anymore. So the first question we have to ask, is the technology a complement or a substitute with the workers?

And you can imagine, just think about our manufacturing example that we talked about. Sometimes technology is a complement with the R&D workers in the manufacturing. So they are the ones who are now going to be able to do newer designs than they would have before, and so they are more productive with the technology. But the line workers might get displaced because the robot comes in and does the riveting or the bolt tightening for them, and they don't need as much of the workers.

The same when the tractor was a substitute for the workers in the farming sector. So question one, is the technology a complement or substitute with the workers in the economy? Questions two, if the technology is a substitute with the worker in the economy, how easy is it for those workers to get reallocated to a growing sector? So at a given point in time, any time in the history of time, some sectors are growing and some sectors are shrinking for a variety of reasons, technology being one of them.

and the workers in the technology get displaced in the technology sector, how easy can they be absorbed into other sectors? In the United States, in the UK, in the 1950s and 1960s, they had those telephone operators. This was a big employment sector, particularly for working women at the time. There was approximately zero telephone switchboard operators in the economy now. Those sectors got displaced and those workers got reabsorbed.

And so the question we ask is, in the sectors for which workers are displaced, they have some skills. The sectors that are growing in the economy demand some skills. How big a gap is there between the skills that are provided by those displaced from the shrinking sector and the skills needed in the growing sector? So let's think about the movement from agriculture to manufacturing. My dad's dad, my granddad, was a farmer.

My dad was a manufacturer. And so the skill mix, the skill domains of agriculture and manufacturing aren't that different for me. So there is some difference, but they aren't different to them. Both of them have a manual component to them. Both of them have some kind of routineness to them. And so...

When the workers left the farm sector, the manufacturing sector was growing and the skill mismatch was relatively small. It wasn't zero. And there was some adjustment that needed. The farms tended to be in the countryside, the manufacturing were in cities. So I don't want to say it was a seamless transition, but the skill step size was smaller. Now the manufacturing sector is shrinking and we're growing in cognitive services around the world.

And those cognitive services, the skill step size is a little bit larger. My father, displaced from a manufacturing sector, wouldn't be able to easily transform into a hospital orderly, or a data scientist, or any of the other kind of growing sectors, even conditional on a given level of schooling. So the skill gap is relatively large. So, one, is the technology complementary substitutes with workers?

Two, if the technology is a substitute with workers, how easy is it for those displaced workers to be reabsorbed into the economy? And three, are there any barriers to that adjustment of being reabsorbed? And so if we need to get skills, if the economy used to be manufacturing that had a certain level of skills, and now the economy is cognitive circumstances that have another level of skills, and there's a skill mismatch today,

How easy is it for the workers to accumulate the skills in the new sector? And if it's not easy, is there a role for policy to help with that transition? And so when we're asking, are the robots going to come and take your jobs? The answer is, for some of you, it will. And that's the way it's been for centuries. The question you should ask is, how easy is it going to be for you to be reabsorbed into other sectors when those jobs go away? And that is going to be...

kind of the discussion around the policy that we should be thinking about. And a lot of my research, again, in economics, again, we're very geeky. We'll have phrases like everything's elastic in the long run. That means all adjustments will eventually take place. And a lot of my research now is how long is the long run? How long do these dynamics take in terms of adjustment? Because that is really going to inform kind of how our policy responds in terms of the labor market. That's all I got right now. Thank you.

Thank you so much for an absolutely fascinating presentation, Eric. So I'm now going to invite responses from our panel, beginning with Professor Chrysanthi.

Well, that's what it says here. Sorry, Chris Anthony, but to give you a shot there. So just to remind you, Chris Anthony is Professor of Information Systems at our Department of Management. Over to you, Chris Anthony. Thank you, Sarah.

Back in the 1980s, early 1980s, a friend of mine who was doing a PhD in artificial intelligence was invited to participate in a panel discussion in a think tank, a major prestigious think tank here in London.

to discuss a policy for preparing for a society with little need for work and therefore lots of leisure. And there were policy suggestions such as the school curricula must change to prepare people for fulfilling lives without having to work a lot.

and, for example, sports, arts, so that they have fulfilling things to do. In the following decades I haven't noticed that much less work for myself or for others. Occasionally there was unemployment, but...

It came in waves and yes it was linked with waves of new technologies but nothing very drastic. Now you may say this AI this time is different and indeed the machine learning technologies and the LLMs today are very different from the expert systems that were developed in the 80s.

Still, this anecdote does suggest a few things which we understood by studies of digital innovation in the following decades since the 80s.

So, I will mention three things that we learned from studying technology innovation. One is that it is predictions on the basis of what technologies are designed to do, like the machine learning and the LLMs, is very likely to lead to wrong predictions. It's very likely that these predictions are not realized, are wrong.

And that's because it is not that technologies that determine socio-economic effects. It is lots of other factors such as how management reorganizes work and the culture of the organizations and the societies where innovation occurs and of course politics.

All of these things are very important and we have to take them into account when we try to understand what is happening with digital innovation and work. The other lesson that we learned is that although there is very often when we have new waves of technology a sense of urgency, actually it takes time.

It takes a long time for technologies to mature and for organizations to readjust, to change either by design or by practice their work processes, the business models to achieve the promised benefits from technology. And probably in that time there are opportunities to mitigate some of the

negative potential effects. And third and very importantly, what we have experienced is multiple technologies interweaving with each other to create technology, digital technology infrastructure. So it is very difficult to

to predict on the basis of what a new technology can do, no matter how potentially transformative can it be, nevertheless on its own it's not going to lead to effects on work availability and what kind of work. We need to understand the whole digital transformation that happens

with infrastructure of technologies to be able to understand what happens with what. And the good example of that is e-commerce and the way it has transformed retailing. So, to your questions.

Will technologies displace the new technologies, AI? Is it displacing and continuing to displace jobs? Lots and lots of tasks will be automated. They are being automated as we speak, they continue to be automated.

but that's not a good basis for prediction. And we have to see how organizations will restructure themselves with the destruction processes. So what happened in e-commerce, for example? Major changes, and yes, lots of jobs were displaced. Shop assistants, for example. But many new jobs were created, both in the administration of e-commerce, but also manual work, for example, in stocking,

workhouses and drivers for delivery, let alone cyber security, privacy and lots of technology jobs. And therefore, coming to your next question, is existing labour reskillable for the new jobs?

quite a lot of them are, particularly as the digital native generation is becoming a big proportion of the workforce and for them it's quite easy to

take digital skills and therefore either complement their jobs with new technology or to move in jobs that are very digitally intensive. But there is another change which is very important and I want to bring to your attention, say in the e-commerce, that quite a lot, a very significant change is change of work contracts.

It is that we have the gig work, for example. So work which is project-based, precarious, not very secure by people who are not members of the organisation. And this has implications for your third question. Because individuals bear the costs for re-skilling.

rather than in the traditional industrial and post-industrial era of corporations where careers were fostered, reskilling was happening and therefore it was not so very dramatic and not so financially insecure for people moving from one type of job to another.

So altogether, I don't expect mass unemployment from the latest type of AI and wave of digitalization.

But I think we do need policies and business leadership which will pay attention to the quality of the jobs available. And this is for me the biggest thing at stake, the work contracts and the kind of work that is available. Thank you so much, Chrysanthi.

I'm now going to invite Noam Yachtman, just to remind you, Drummond Professor of Political Economy at the University of Oxford, to respond to Chrysanthi's actually hopeful vision of the March of the Robots. Yeah, so I'll add a bit of hope in responding to Eric's question also about...

sort of adjustment and elasticities along the economic margin, where do we see occupations and particular skills that AI may struggle to be a substitute? So to me, I see three categories, and there are probably others, but one is...

empathy-intensive positions, and I think you can imagine healthcare workers and other services. I think doctors by now are not as good at

diagnosing or making prescriptions as algorithms, but we still prefer human doctors. Maybe we prefer human doctors with algorithms. Another area is what I would say is responsibility-intensive occupations. Again, I think algorithms probably do a better job than judges in

in making decisions about whom to grant bail to. But we still often, I think, would prefer to have a human who takes accountability or is responsible for some of these decisions.

that might become even more crucial, even as AI gets very, very good in the form of large language models, where these models are incredibly powerful. But I think we all know they can hallucinate. And so you always want to have an expert

who is truly an expert, monitor the output of AI, and especially sophisticated AI, that might be particularly tough for a non-expert to monitor and evaluate the difference between a very sophisticated hallucination and the truth.

So the third category I would say is performance intensive occupations. And those are occupations where we care about a human activity per se. And that might be music, the arts, athletics,

where again I think for now we prefer seeing Lionel Messi play as a human rather than watching an AI Lionel Messi on FIFA that may change across cohorts I don't know what a 10 year old would say that's a bit frightening but

So I think there are categories of occupations, and not all of these are particularly skilled. Some of them depend a lot on the structure of work, as Chrysanthi said. So what are the jobs where we demand human accountability and responsibility? That might be determined by regulatory policies. It might be determined by management. It might be determined, in some cases, by litigation and litigation risk. And I think that's incredibly important.

I would also say there are some real opportunities coming from complementarity. So I think in many Western rich countries, but also Eastern rich countries like Japan, we worry a lot about demographic change and the aging of the population.

one could imagine directing AI to complement, let's say, older workers to keep them in the workforce longer. And that might help us address these types of demographic concerns. So I think there is opportunity from elasticity on this economic margin. I think where I see more risk is along other margins of adjustment.

So, Chrysanthi mentioned this question of a leisure society and Eric has some really, I think, provocative and fascinating work suggesting that video game technology has been an attractive use of time, especially for young men, as an alternative to employment.

while while there hasn't been a leisure society one could imagine a group of people getting larger and larger who are pulled toward an ai powered metaverse which is one potential adjustment uh in the long run at least for a subset of the population um the other big one and given my research i think one that that i i think about and worry about most

is the margin of political adjustment. And when we see historically the displacement of workers, it often comes along with changes in the concentration of wealth, changes in firm market power. And when we look at the early 20th century, we see a massive increase in inequality. We see concerns about concentration of market power, just as we see today. And those types of concerns

although eventually we found jobs for workers, the political tensions of that time were by no means sort of predetermined to evolve in a way that led to liberal democracy. Things could have gone the other way. There was a lot of violence. There was a lot of contestation. And I think that's the margin of adjustment that I think is particularly concerning to me. Thanks. Thank you, Noam. Thank you.

Hi, I'm interrupting this event to tell you about another awesome LSE podcast that we think you'd enjoy. LSE IQ asks social scientists and other experts to answer one intelligent question. Like, why do people believe in conspiracy theories? Or, can we afford the super rich? Come check us out. Just search for LSE IQ wherever you get your podcasts.

Now, back to the event. Just before I open to questions, I would just say that I really strongly agree with Noam's last point. Having studied Russia in the 1990s, a brutal decade of economic adjustment, and we see the political impact of that now and kind of, you know, attraction to authoritarian leaders. And, well, we're living with the military consequences of that.

So now I'm going to open the floor to questions from the audience, both here and online, but first of all here. As I said earlier, if you're online, please type short questions into the Q&A box.

But if you're here, raise your hand and wait to be called on and don't speak until the microphone arrives. So please include your name, affiliation and kindly confine yourself to one short question so that we get as many questions as possible. And I'm sure all our panelists will have a chance to respond to your questions. So who's going to give me my first question?

Oh, right. Well, that's a startling phrase. First of all, I'm going to go to the person there in the red top. You? Yeah. And then I'll...

Fantastic, thank you. Hi Rick, thank you Sarah. That was really interesting. My name is Subhashini, I work in digital marketing and I also study organisation behaviour. So the panel and the discussion today seemed really intriguing to me. My question is specifically on what Noor mentioned last and Chrysanthi also mentioned in between.

there's a big component of gig workers, of unregulated workers, even in the creative industries in the field. And the creative industries is where applications of AI are being fruitfully used, but at the same time also creating more of a wealth gap. And as you rightly mentioned, the upskilling is dependent on the creative workers themselves. They are more responsible for it

And so I wanted to ask what kind of policies can we see coming up? Or we may have a case practice from the history of US and UK, where we can bridge the wealth gap in these populations, specifically in the creative populations, where they themselves are right now responsible for their upskilling.

Well, you start with the political. The legal part is you. So then I'll talk economics. Sure. So I think on the political front, I think some amount of organization by workers seems important. Some ability to coordinate, whether it's unions, whether it's at a level higher than unions like political parties, some amount of representation

just seems critically important to reallocate the rents that are created. And there are rents created, as you say. There can be a flowering of creativity right now. It seems like that flowering of creativity may lead to very skewed returns, and you can imagine some sort of social contract that reallocates those returns. I would say these types of organizations...

political organizations, unions, will sometimes be able to act within sort of institutional frameworks. I think history also teaches us sometimes that they act outside of institutional frameworks. And so then that leads to things like strikes, which sort of inside and outside institutional frameworks, but also protest movements, social movements that I think we'll very likely see in the years ahead.

Yeah, Eric and then Chrysanthi. I'll just be brief on that. So we know historically there's an any group that has a little bit of power, they're going to want to exercise that power. And so firms...

Try to exercise power. In the literature now, there's a large literature, some of it kind of right here in this building, thinking about market power on the part of firms. We call that monopsony in the labor market. So how much power do firms have over workers to suppress our wages below our marginal product? Our marginal product is how much we would be worth, and then the firms might want to pay us less than that. Now, why can they do that?

They might, we're trapped in the short run. If we could just go and the market could work and we could move where we need to move, we might be able to undo that through our mobility. But if mobility options are limited, I live in a town, there's one firm, that one firm has market power over me, I'm kind of trapped. And so unionization kind of born out of that in the manufacturing sector to try to undo some of that market power.

And so the strikes are one way to kind of help equilibrate some of that market power. And so, you know, there's benefits and costs on both sides. Both sides want to extract rents. If either side extracts too much rents, that could have negative productivity effects for the economy. But you might need some sort of, you know, coordinated effort. And, you know, no one thinks a lot about, you know, regulation and legal, you know, how we think about, you know,

firms wanting to get around the regulation and maybe what is the optimal response of regulation to that. Part of that is to kind of keep market power. The economy works well when there's not too much market power on either side. And so how do we get that balance? So the gig workers kind of is going on now. It's a sector that grew very fast. And if we're just thinking about just Uber and taxis had a lot of market power before. Uber comes along and does that, but then the workers...

you know, we're kind of left on the sideline. And so how do we kind of rebalance that? I think that's kind of a, you know, part of the transition dynamics through to some of these new technologies. Yeah, a few ideas come to my mind. One is that

Unfortunately, unionization doesn't quite work for the gig economy. That's one of the problems. Not all work in the gig economy, of course, is not good. There is scope for entrepreneurial initiative. Many people find opportunities.

Nevertheless, there is a need for continuous reskilling. And who will do that and how will it be done? Government policy has, and government initiatives have a role to play here. There is this concept of lifelong learning. Unfortunately, I don't think it has been implemented in adequate way in any of the societies I know. So it could be improved, it could become more effective. But also business leadership have something to do here.

I remember in work that I did in around 2010 in China and I studied of all things in the world, Taobao that introduced at that time e-commerce in a big way.

And Alibaba, Taobao, the corporation, had a big role to play in creating the skills that were needed for their own business. So they had colleges training micro-entrepreneurs to do business on the Taobao and to do their own business management and accounting, etc., etc.,

Therefore, we could have initiatives from the industry, and it is good for them to establish themselves, which benefit and create the workforce that is needed for them to flourish as well.

This can be encouraged by policy. I just, as Sarah says, want to bring some positive views of what might happen with these technologies that have created fear in terms of the potential consequences that may have on work.

Thank you, all of you. I'm actually really happy to see that industrial relations and unions, which is my area, are getting quite a look in here, although I'm not super optimistic in that unions are declining worldwide and smaller workplaces are harder to unionise and authoritarian countries make it very difficult for unions to organise. But anyway, I will not monopolise my position as chair. We had a...

I have a question there, Raj. Go ahead. Oh, okay. It's gone to the person right behind you and then we'll come to you. Thank you. I'm from Oxford SDE, it's a research organization, and the question is if you are thinking in terms of more of circular economy where the people can bring on board

the economy, use the economy, bring the innovation at territorial level. I think this is a way where they can grow the territory, can grow the organisation and it's every border benefit. That's the question is what do you think about this? So I think that's a question about your views about the potential of the circular economy, am I right? Circular, not linear. Yes, circular, yes, thank you.

I don't know where to go exactly with it. So technology has effects on lots of different dimensions. And so there are some that we've all been talking about, that there is consumer surplus that comes from new technologies that creates spending, that spending creates employment. That kind of things work. So I think we're all on the same page, that we've gone through periods of time where sectors disappear and life goes on. And so the question is about the reallocation.

And so, as Chrysanthi was given her remarks, she believes the reallocation is going to happen quickly. It's going to happen quickly because the workers are going to have the skills to move or new jobs are going to be created to move into those jobs. And that's a world that is really good. And so the question becomes what we should be concerned about is if we ever see that there are barriers to that adjustment.

And I think we all kind of come from the same side that things are elastic. And so we've gone through these things in the past. New jobs will come out, new demands for skills, new people will innovate, that innovation will create some resources, those resources will be spent and that they spend on jobs. All of that's just the nature of an economy.

So the question is, is there frictions? And so one friction is a potential skill mismatch that occurs. And one, again, the skill mismatches tend to occur if sometimes the technology occurs very quickly. So if there's a lot of change all at once and jobs change very fast, it might be harder to adjust. But if this is kind of a linear trend that's happening over time, you know, the people who acquire skills change.

will then move to the parts where the skills are going to be needed. And that's been, again, happening for centuries over time. So it's the barriers that we need to think about to those skill adjustments when we start talking about policy. Speed of adjustment might be one thing that we could be thinking about. When things happen quickly, the adjustments might take a little bit longer to occur.

My position is not or my idea is not so much that the adjustment happens quickly, but as I said, that the very influential job destroying changes take time to occur. In that respect, there may be, for example, enough time for a generation of workers to come into the industry with new skills and very skilling to happen.

It's fundamentally that I don't think that all these effects, drastic effects of the new AI as it is very often projected in the media will happen next year. Already we have news in The Economist two weeks ago where journalists have done their surveys and they tell us,

Actually, the new type of AI is not diffusing, taking up as quickly as we thought. And that's not a surprise. So the speed of adjustment, I think we're all getting it. It's not a difference. It's exactly the same, the speed of adjustment. And so what happened in manufacturing is there was a shock that her concentrated in a very short period of time when that early 2000 period when China did come on and automation was ripe at that moment.

that happened, those adjustments occurred quick. And so you might imagine that has a very different effect than if it was spread out over a decade or so. And so I think, again, I just want to make sure. I think we are saying the same things. Sometimes it sounds like we're coming from different-- But they're coming from different sides. But yeah, they're different sides. But I think the speed of adjustment is an important component to that. So maybe adding one other thing, that the scope of AI's effects

and recent technological change. Also, it can be much broader than earlier automation and robotics in the sense both economically a broader range of jobs potentially, including highly skilled jobs and cognitive jobs, which would change things. It's harder to find a new job if a much broader range of jobs

is being taken by the robot. But more generally also, the fact that AI and new technologies

really do have deep social and political effects. And so one of the things that is important is not just how quickly can I find a job, but do I think that maybe a politician is offering me an alternative to looking to reskill. And I think what we see now is some sort of attempt at a political free lunch.

And those types of politics might also be, in some ways, enhanced by AI, unfortunately, its role in misinformation and things like that. So, yeah, that's another sort of friction. Did that happen in prior periods of adjustment, too? I mean, again, we had newspapers and radios as opposed to Internet,

but did you see those similar patterns when we went through the early 20th century? You do, in fact. So you see the rise of newspapers in urban areas that play a role in pushing sometimes extreme politics and things like that. So that's a very interesting point. Yeah. So question here, and then I'll--

This is Raj Chowdhury from Harvard. So great session. Thanks all of you. Quick question. So Eric, you mentioned inter-regional mobility as a solution. And I totally agree. But we know from the recent work-- there's a paper by Ed Glazer and Larry Summers that inter-regional mobility is declining in the US.

And I wonder whether part of the solution has to be moving that factory worker from Detroit to Kansas where they're building windmills and solar plants, but also across the border to Mexico where they're building these factories and these workers could be. So I think in sum, the moon in the north seems to prevent migration from the south. And I wonder whether we actually need policy to move these displaced workers from the north to the south. - Let's step back and there is

kind of a large literature that on the margin over again, medium run periods of time, long run people tend to move. People migrate over centuries to different places and even decades to similar. But moving is costly, particularly at certain parts of the skill distribution. So a lot where I grew up, I get the town I grew up in upstate New York, very few of us went to college. So I was an outlier relative to many of my peers. And during in those towns, their community is their

social network, it is their insurance network, it is their human capital development because we're taking care of each other's kids when somebody gets sick or goes to the job. And so it is costly to be moving. And so that's even without having to think about crossing across international borders where you might need a new language and new social norms and things of that nature. So while mobility in theory kind of sounds good, you want to move where the jobs are spatially,

it is a cost to get people to adjust. And so, you know,

how much subsidy would you need in order to move? Do you need to move your grandmother or your mother who's taking care of your children at the same time? And so thinking about that as a policy, it's an interesting thing. It's just a complex thing because mobility. Now, again, natural forces, people do tend to move where jobs are over long periods of time. And again, it's the young that tend to be more elastic. It's the young that tend to pick up the new skills in the human capital.

And so maybe the youth kind of responds more to those potential subsidies of mobility than older incumbents.

I would like to mention here that another thing that happened is remote work. And that, to some extent, has limited the need for physically moving. And there are very interesting phenomena emerging there. We still have to understand how post-COVID work

office work will unfold, whether people go to offices or not, but this is another possibility of technology. And it has consequences. It has consequences of where work is available. Very often it goes abroad by being remote. Very skilled programming, for example, the big technology companies in the US, are

appoint programmers all over the world who work 24 hours a day remotely? Yeah, I agree with everything that's been said. I guess I would just say, again, since I'm putting on the political economy hat, that I think history tells us that these types of movements to very new settings, if it was like rural England to the cities to work in the mills,

they often come with social and political challenges as well. And I think it's one thing to sort of think about purely economic incentives and then you reduce mobility costs and people move in response. But yeah, the social and political impact would have to be managed. Fantastic. Now I'm going to take... Okay. Yeah, there. Oh, sorry. Yeah. You in the blue shirt.

Thank you. And then after that I'm going to go, Danny I'm giving you warning, I'm going to come to the online audience in a minute, but yeah, go ahead. Thanks Sarah. My name is Billy, I'm a recent graduate from the Visiting Programme of LSE. My question is, I think several panellists mentioned we need to reskill, we need to train our labour to take up new jobs, but

I'm wondering whether government should take up this responsibility because like several years ago, many governments expect programmers will be in huge demand and they train up many programmers in their elementary education, but with generative AI, actually many programs can be

written by AI. So actually we need some other skills. So it's very difficult for governments to keep up the pace as quick as the commercial sector or the businessmen.

So should government take up the role of directing the retraining or government should only give the incentive to companies to retrain their employees to absorb their labors?

Because that is a very critical question, because we are of that which region will be the global economic driver. So that's it. I had three things that I was hoping would come up in the questions. This is the first one. So when we start thinking about where these kind of potential skill gaps, if there is a barrier to skill adjustment, how could they go through? And so...

I've been thinking quite a bit about this. Apprenticeships are one way where the firms themselves would do some of this training in-house. If there's a skill missing, they could come and train. Now, the problem with apprenticeships sometimes is if Sarah trains Noam and she pays that cost to do those training.

I'm going to sit there and steal Nome. I get a trained worker without having to pay the adjustment costs. So there is a coordination issue that comes with some of these training programs. And that's where, potentially, the government can kind of come in and help with that coordination. So a tax subsidy for kind of training and apprenticeship might be able to get an equilibrium where both Sarah and I are choosing to do the training in-house. And then we get rid of the--

kind of the public goods problem of Sarah paying the cost and I getting the benefits. And so having some of that is, I think, good, but it might need some government help to do. The German model is very good on that dimension. A lot of training is done within the firms and there's an equilibrium where that tends to be supported.

You also touched on another kind of broader component. Again, I want to be quick because I know there's lots of hands out there that have been up there for quite a bit. But when we're starting to think about public education broadly, and so back in the States when I was growing up, there was a lot of vocational training done within the schools.

And so at this time, there was a chunk of my classmates when I was doing calculus that was going off and kind of learning, you know, kind of, you know, motor repair, auto repair, and other types of tradable skills. We tended to move away from that for a little period of time for exactly the reason that you mentioned, that skills tend to deplete. And if I teach you how to do something today, maybe you don't. That skill that I'm teaching today kind of goes away. And so there was a movement of training, trying to teach you how to be flexible,

I kind of like the old method because getting some skills, getting you into the labor market allows you then to accumulate those skills because some of these adjustments occur over time. You could slowly kind of learn them. So I am also thinking there's a role to be thinking how we kind of, you know, do secondary kind of education where we could try to teach some skills, vocational skills that the labor market needs, you know,

as part of the policy. Again, I know there's lots of questions, so I'm trying to be brief, but I ramble. I apologize. I agree with--

Oh, I think so. One small institutional note. In 19th century Britain, you had master and servant contracts which bound workers to their employers. They're functionally a bit like non-competes today, and so you could have freedom of contract into these sorts of apprenticeship arrangements. Yeah, freedom of contract. I'm looking forward to you running on that platform. But there might be other ways we might be able to do that to try to get that.

Right, Danny, your moment to shine. You're going to read out some online questions. Amazing, thank you. Why don't you give us a couple? I'll give you a couple from online, yes. Thank you. So we have one from Antonio Santos Spirito, an LSE visitor from Italy, who asks, as unionization outside of manufacturing is more difficult, can you see the base for income distribution being reconsidered, such as minimum income for all to control the level of inequality? Yes.

David Walter, alumni from Birkbeck, London asks: "As there was in the early times of the Industrial Revolution in the 19th century, with the impact of the Luddite movement, can you see any impact of a modern movement regarding an over-reliance on technology?"

You start, and then we'll kind of work it out. And also UBI, universal basic income, as I understand the first question. I'll talk about that one. No, I mean, I think as I suggested, I think, and so Chrysanthi mentioned, unions may not be the easy solution given the gig economy, and as Eric said, you know, I think this is,

Organization within the firm is one possibility, within institutions is one possibility. I think we are likely to see social movements outside the firm. That's what the Luddite movement was. Eventually it did lead to policy changes at a higher scale, not very directly but indirectly. One of those might be a push toward rethinking the social contract.

And with the policy, we talked about before, again, what is the friction is the key question. So you should try to figure out what the friction is. So there are certain policies out there, like universal basic income, which you could help people, but what is that going to friction? Because there's going to be incentive effects. If I pay, you know, Crescenti

some money that might make her adjustments a little bit less. So we've got to figure out what is preventing, because if it's a liquidity constraint, maybe she needs money to take care of her kids while she is going to get schooling. We should think about that as the friction and the money could help with that in other policies as well. So I think it's always good when we're thinking about policy to go to my third point. What is the friction that is preventing the adjustment from occurring? And then how could we help with alleviating that friction?

Fantastic. Right, we'll have another in the hall question. So there please with the white shirt and blue waistcoat. And the gentleman with the green has his hand up the first time all the way through. Which one's that? Right there. Oh, sorry, Simon. I know you're like in my vision, but disguised by someone who didn't see your hand. I'm sorry. It was a

I was a business practitioner. My question is that

A few weeks ago Martin Wolf from the FT wrote a piece and he said we must socialize AI and if we fail to then our working lives would become like a latter-day feudal system. I was wondering if the data you showed Professor Hurst actually was an early sign

of a feudal future for the workforce. And so I'm going to use that as my launching off point for the second thing I hope we talked about, which is kind of the distributional aspect of some of these policies. And again, the feudal system has some of that component to it. A lot of the stories that we're talking about, we tend to look in the mirror and think about us.

I don't worry about us in this room too much in the sense that we tend to be elastic to potential shocks. And so the question is, as these technologies are coming, is it going to be concentrating power, resources, employment because of the complementarity and the rents

into the hands of a set of people, and then there's going to be another set that's going to be potentially left behind. And so that's why I'm not always worried about the computer programmers at the top, or I'm going to be worried about if somebody comes in and AIs me away, and you can have a virtual lecture from me from some machine, it'll probably be pretty good.

I'll be fine. I'll figure out a way. I'll make that robot or something. I mean, so the question is, how do we figure out the adjustment at different levels in that inequality part? And I think I want Nome to kind of come in, because usually it's not the technological disruption per se. It is the technological disruption in exactly the way that you kind of hypothesize that it's going to be some winners and lots of losers. And those lots of losers are going to be the ones who are going to, you know,

I want to ask Noam this question. We tend to see, over time, waves of populism.

I want to see those waves of populism, my gut is, and you would know better, seem to be associated with these kinds of technological disruptions. And how do we think about that in protecting institutions as a component of that inequality? I think you're spot on. It's the inequality component of all of this that I think is going to be first order. So I totally agree. I think that that

the adjustment process that as economists we might hope would happen and might be aided in some ways could be abruptly stopped.

by some sort of large-scale political disruption. And I think there are people who already feel economically disrupted who could be politically organized. I think that becomes more likely as more people get displaced. But again, it becomes more likely as AI is used as a political tool as well, which is what's so strange about this time. So the earlier industrial revolution

involved the concentration of people in urban areas there were technologies like radio newspapers that were also used for misinformation so this would not be the first time but ai might be particularly powerful um and and i i think this is a huge risk so what what ends a feudal system um you know

major shocks. So the Black Death in 1348 puts an end to feudalism or weakens it. Unfortunately, I think sometimes it takes large-scale disruptions to lead to a rewriting of a social contract. So socializing AI is a bit of-- I mean, it's a vague policy prescription. But I think the point that I think we need to think about is

the concentration of political power. And so I think if what we mean by socializing AI is that right now some companies or individuals that have control of major platforms or major data sets also use that to acquire major political power,

that can be a serious problem. So I think what I would do is try to enhance the institutions of liberal democracy, freedom of the press, freedom of association, education, not just education for skills, but also education for civic society. I think that becomes crucial. Can I ask a normal question just quickly? Do you see the view of the rise of populism

kind of globally with a common cause? And is this part of the common cause? I mean, because it seems like it's a correlated movement across. Even though we're being recorded, and so of course that always puts you at risk of everybody taking... Sorry to ask that. I really thought that question. No, no, no. Absolutely. So I think, to be clear, I don't think it's a monocausal phenomenon.

So it's not just to say, you know, there is technological change or there is inequality, therefore there's populism. But I think it's clearly part of what makes populism an appealing sort of framing from the political side. Absolutely.

Chrysanthi. I think economic inequality of the grand scale is unfortunately already entrenched globally. And part of that is the industrial concentration.

This is something that we have large companies that are very specifically geographically concentrated with huge consequences for economic inequalities around the world. So most countries become just consumers of the products and the services

or a few very large tech companies and that does require regulation. And of course it does require coping with the anger of those who are left behind and they have to live with very little when we have obscene wealth concentrated in the hands of a few people.

Yeah, absolutely. Right. Another question from the hall. So, yeah, you that I called on and then you didn't get your chance. Thank you very much. I'm Bakhtiar, a graduate student at the LSE School of Public Policy. I was particularly excited when I saw the name of the event because I'm characterizing my dissertation on the impact of AI and automation on labor markets in Azerbaijan. And my question is about what can policymakers do

suggest to adopt to ensure that the integration of automation fosters inclusive economic and social progress rather than deepening work force disparities and leading to inequalities? Very important question. Do you want to start? What policies, you said, what policies can address the inequalities? It seems to me that... And inclusive. To create, yes, inclusive societies.

training, retraining of the workforce to make sure that they can effectively participate in the new available jobs. Legal regulatory measures so that there are redistributed effects and there is no concentration of wealth in specific companies and specific regions are two things that come to my mind.

Sure. I mean, one thing that comes to mind to me is trying to generate some understanding of what would be labor complementary AI. So I think there's some evidence from manufacturing in Japan that in Japan, so thinking about what Chrysanthi said about organizations that have a particular approach or structure to manufacturing, if you talk to manufacturing workers...

in this case, not manufacturing workers, but workers kind of across the spectrum, say, what would you use AI for that would make you more productive? I think that just generating that sort of knowledge, I think, would be extremely valuable. That could be

be fed into the private sector? The current early evidence, actually, is that AI is used mostly for augmentation rather than substitution. And it is used, for example, to enable low-skilled people to do jobs that require higher skill. So it has this kind of

interesting effect of upskilling existing low-skilled workers to be able to do more demanding jobs, hopefully to have better incomes as a result of that as well. But that's anecdotal evidence at this stage. It's early days. And technologies, you say that they have all these problems and et cetera. We need to... But technology will improve. Can you go?

One of the problems that we had with digital innovation over the years is our inability to foresee the big changes. We got it wrong again and again. We thought Internet will bring democratization. We thought that will bring inequality. And we ended up with the opposite. So we try at the moment to discuss about the foreseeable future, but the foreseeable future is very short. And one thing I expect is surprises. And I don't know what this will be.

Eric, last word on this very interesting question. I just want to underline something that I think Cassanthe has been talking about and known such a

the concentration both in the labor market and the product market of big firms, of these scalable, if you're out there, is something that I think we should be pondering. Is that good for productivity, having a big firm that could maybe lower costs, or is that bad in the sense that they are concentrated in exercise rents? I just want to underline this kind of theme that has been touched on because that is something, again, that policy might be able to think about.

Right, I'm going to get another online question in case our online audience is getting frustrated.

We have a few anonymous questions, unfortunately, so I'll read those out. We have one which asks, do you think job polarization will shift the Phillips curve and make monetary policy measures more difficult? That was from my mom. We also have a couple of questions generally around the theme of automation

and AI in teaching and education and how that's going to affect graduates going into classes and their fears around job prospects being automated and kind of the effect within teaching itself such as automation in essay writing or assessment.

I'll do the first one very, very quickly. So again, the question about policy, like short-run stabilization policy, that's a Phillips curve, the trade-off between unemployment and inflation. There's nothing in inequality that I worry about that. That is a short-run phenomenon. Inequality is a long-run response. If central bankers get it wrong by two months versus one month, life will be fine. And so, yeah, there's nothing about that that we should be worried about.

Mom. I'm glad you answered that question. So I'm going to leave the others to you guys. A few words about AI and digital technologies in education. As I said, the important thing is that work is redesigned and readjusted with the technology.

We will all adjust. We will adjust our assessment, we'll adjust our teaching. A few years ago we had this huge scare that the MOOCs will take our jobs. It hasn't happened. So I think that... I hope that it will enrich our jobs, complement rather than somehow make us impoverish our jobs or take our jobs in academia and in schools.

I agree. I think teaching is one of those relationships where some component of the human interaction for now especially remains paramount. And I think AI will complement that. Yeah, exactly. And our presence of you in the audience tonight shows that you have that hunger. You're not all sitting on your computers. So, yes. Back in the white jumper. Yeah.

That's you, yeah. Now just as a statement while we wait for the question, imagine if you feed all of these questions into ChatGPT and compare the answers ChatGPT gives you to our responses. Are you saying that's good or bad? I don't know how I would do. It would be interesting to know. What would be more interesting, us or-- You'd definitely be more interesting. Thank you.

Actually, I did ask chat GPT in preparation. I was curious to see what the replies would be. And sometimes they were trivial and sometimes they directed my attention to things that I needed to remind it of or simply I didn't know. Very good. So...

My name is Alec. I'm an exchange student from George Washington University. My question is about AI, fusion, and quantum. So assuming that in the medium term all of these

are possible to create and implement at large scale, how do you think societies and labor markets would react to a simultaneous dramatic increase in thinking power in terms of AI, energy by fusion, and then computing power through quantum?

Again, I am so boring sometimes because everything comes down to those three things for me, whether technology is going to be complements or substitutes with the workers. Some of us it's going to be complements. Some of us it's going to be substitutes. It's going to be then what affects the adjustment of those workers to the sectors that will be growing, just like the robots came and killed the agricultural sector. Again, we didn't call it a robot, but it's exactly the same thing. There was some technology that displaced a whole bunch of workers. And the speed of adjustment could be one thing that affects workers

the speed of the changes could affect the ability to adjust. And the third thing is, is there barriers to adjustments that wouldn't occur on their own that policymakers should come in and do things in? Those things then lead ourselves to whether the firms are going to do some adjustment, whether the policymakers are going to do some adjustments in response to that. So I don't think any of these are special.

when you kind of do that. Now, they might be special in their magnitude. They might be special in whether they hit a certain group that allows that group to be more adjustable. Again, I don't worry about the high skilled part of AI coming in and taking my job. I'll be fine.

And so it's really how substitutable or complement or how easy the adjustment is. And then are there barriers to adjustment, regardless of what that is? So 60 years from now, there's going to be something else going on. And you're going to have a quote from me. Oh, my God, that ancient person was worried about this, you know, you know, back in 2020. And again, this will happen for centuries. These things have been happening.

Chris, something? Maybe I'm prominent enough to get a quote, I don't know. I find interesting positive spin that we are going to have the fusion technology energy and we are going to have unlimited computing power, etc., etc., and of course there's very powerful AI. It will take time, years, decades, and you know, I'm not quite sure it will happen.

or what will happen in between so that to be able at this stage to predict the effects. As far as digital innovation is concerned and radical technologies, yes, the change is slow and it is difficult to predict that many years ahead.

Yeah, I would also say Robert Gordon has this really nice book on the history of American technological change over 100 plus years. And he makes a point a few times of saying there are certain changes that could only happen once. So electrification and some of the consequences, the discovery of the spectrum in radio and communication. And I think there might be truth to that.

These were absolutely groundbreaking technologies, but even they diffused slowly. And so I want to sort of reiterate that. And I think quantum computing and fusion energy have a feeling of being a qualitatively different approach to solving certain problems. And we might eventually say that could only happen once, but even if we do, probably the diffusion of the technology and the impact would be felt over time, and these same frictions will become relevant.

Okay, right. I think we've got time for two last questions and I'll take them both and then we'll conclude. So, at the back with glasses and in a stripy jumper.

Thank you for this wonderful speech. I'm a master student in economics from LSE. So I have two concerns about the technology development on today's economy. One is that when Professor mentioned that the technology can be a substitute, a complement of the worker. So I'm concerned about what if the productivity increase

but there is a fixed amount in the market. So even if there are a lot of firms that can produce more stuff with the automation, but the consumers, they don't need so many stuff. So is that possible to have competition so that only several firms can survive in the market so we still have unemployment or a waste of resources from the increase of the productivity?

So my example is like suppose there are so many coffee shops and every day with the coffee machine I can, the workers can produce more coffee every day but like I only need one or two cup of coffees every day so this fixed amount

maybe could be a concern for the increase in productivity. And the related question is that... We haven't got much time, so could you just get to your question? Sorry to interrupt you. Oh, OK. Sorry about that. Yeah, OK. The related concern about the fixed-demand, least-statiated economy is that when people have more income, will they care more about their leisure rather than they want to increase their consumption or they want to work more? So, yeah.

Let's take one more question and then I'll get your final concluding remarks. So here in the blue jumper. Wait for the microphone. He's coming, don't worry.

And I'm really sorry to everyone who didn't get a chance to ask your question. We will have drinks afterwards. Okay, my name is Bhanu. I'm from Indonesia and studying PhD in Southampton. I want to talk about, because many people like talking about the effects of automation and AI for the corporate.

But I want to ask your views on the effects of small enterprise because the companies like to complement for this technology is very limited for the small enterprise. And then also like many function or jobs that can be provided by small enterprise is replaced by the AI, especially in service.

I want to know your views about how this one enterprise can survive with this technological development. Yeah. Okay. So we'll be very quick. Again, we will be around after. My answers get better with drinks. So if anybody wants to hang for that. So let's go quick because I know we're short on time. The market tends to work better.

pretty good for some of these questions. So when we went back again a century, people were worried, you know, there's so much productivity in farming now, how is people going to eat all this food and how the, you know, there's other prices that are out there that kind of guide people towards the right outcomes. So again, prices will adjust.

If there's too many goods relative to demand, prices go down, we're all better off. So we eat some more on the margin, but some of those kind of things come on. With the question of work leisure trade-off, that is an interesting question. I should have been more specific. The Keynes quote that I started was exactly that, that we were going to be so rich that we're going to be choosing to take leisure. I kind of put it in a way that it made it look like it was going to be kind of just a bad thing. But he was writing it in the sense that's not a bad thing, that people are going to be freeing up for exactly like your friend, you know,

you know, 20 years ago. And so there is this trade-- more than 20 years ago. And so, you know, this trade-off is there. But, you know, people also like to work at some point. And they like leisure. And we tend to seem to manage that balance. So the key thing is markets tend to work. When they don't work, that's going to be the role for where policy needs to get in.

I'm happy to let Eric have the last word as he's our guest. I'm also around for drinks, so happy to talk more. Thanks, sir. Okay. So I think I can speak for everyone when I say that this was a really stimulating and illuminating evening. And I think hopeful rather than too scary.

So, if you enjoyed this evening, I'm just going to do a little advert. Please note on Thursday the 3rd of April, Rocco Machiavello, who's actually here in the back, I think, another professor in our Department of Management, he's going to give us a fantastic talk on the ins and outs of sustainable supply chains. So that's Thursday the 3rd of April.

back to this evening i hope that you can join us as we've already advertised for drinks outside thank you so much for joining us today and please join me in giving our three fantastic panelists

Thank you for listening. You can subscribe to the LSE Events podcast on your favourite podcast app and help other listeners discover us by leaving a review. Visit lse.ac.uk forward slash events to find out what's on next. We hope you join us at another LSE Events soon.