Skills in the age of AI
01:30:41 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. Welcome everybody to this event, Skills in the Age of AI, where we're going to be having a discussion about how AI is changing the world of work or potentially changing the world of work. And thinking about how

we might try to unlock the potential of AI while we're still protecting workers' well-being at the same time. So for people who haven't encountered me before, my name is Steve Machin. I'm Director of the Centre for Economic Performance and I'm a Professor in the Economics Department here at LSE.

So I'm very pleased to be here today to chair this seminar. We're going to be discussing potentially seismic changes. Well, there's seismic changes to some, and there may be less seismic changes to other, and we'll think about that as we kind of go through the

and hear what our speakers have got to say, and the kind of changes we're kind of facing in our working lives now. So we have two highly eminent, fascinating speakers who are going to tell us stuff about these kind of questions. So first of all, we have Nobel Prize winner Christopher Pissaridis, who is a colleague of mine, a professor in the Economics Department here at LSE, and he's co-founder of the Institute for the Future of Work.

And we have Mary O'Mahony, who's a professor of applied economics at King's Business School. She's research director of the Productivity Institute, and she's an affiliate with the Economic Statistics Centre of Excellence, ESCO.

So just to say a little bit about both of them before we begin. So Chris has been leading this quite important three-year review, I think, into how automation technologies have been transforming work or have the capabilities of transforming work, society and the economy. And, you know, in some ways comparable to many episodes of economic history dating back at least as far as the Industrial Revolution.

The Pissaridis Review has not only looked at work, but it's also importantly looked at well-being at the same time. So the research is explicitly aimed to try and find out what technology

New technologies, specifically AI, but even broader than that as well, mean for access to jobs, working conditions and employees health. The final report I think was published a bit earlier this year. I don't know which month. January. January. A lot earlier this year.

So Mary's research interests include the measurement of productivity, technology and economic growth and explaining different international differences in those economic measures. She's carried out highly influential work which has impacted the way in which the Office for National Statistics measures human capital. Her research is

including some of the work that's actually been undertaken at the Productivity Institute, has in particular, and I think we're going to hear about some of this today, highlighted the important role of workforce training. In particular, not just training per se, in particular the distributional consequences of that, in particular who gets it and how people acquire skills via training. Okay, so let me just do a few little...

make a couple of points about the format of the event. So where we're going to go is for this event, Chris and Mary will talk each for about 25 minutes, and then we'll throw the discussion open to questions from the audience, both here and we have an online audience, and so we'll be taking questions that Martin will transmit over from the online part.

I should ask you, and I should probably ask myself, to please put your phones on silent so as not to disrupt the event. This event is being recorded and hopefully will actually be made available as a podcast as long as we have no technical difficulties of preventing that. That would be a little bit ironic given the content of the talk if we're not able to transform it into a podcast.

So if you are online, and I think people are online, please submit your questions through the Q&A function that's available on there.

If you're a Twitter, so out of date, or X user, the hashtag is #LSEEvents, and that's LSE and the E of events in capitals. I'm not sure if it's capital sensitive or not. So I think that's most of the organisational things. So let's begin. So I'll hand over to Chris to begin with his presentation. Thank you. Thank you very much and a very warm welcome from me as well in this

a place that I know very well, both as a student, as a young lecturer, and as an older lecturer, and as an even lower presenter. And I will talk to you today about AI. Most of what I will say derives from the study that we just completed, a three-year study that was funded by the Nuffield Foundation. And what I plan to do is to give you

and introduction to the way that we're looking at AI. I mean technology in general, but AI specifically here and how it's affecting labor markets and employment. And that will naturally lead into more specific skills for these that Mary will talk about later on. So we know that jobs are changing and

If you ask what's the reason for change, then usually the reason is technology. The current technology is AI. What do these technologies do? Well, primarily they raise output and living standards. That's what you usually study, the students amongst you, at least when you're doing growth theory as part of your economics. But what I'm mainly...

more than this actually, what I'm interested in, what I've always been interested in my research in fact, is more the structural change that technology is bringing to the economy. In fact, growth comes because of the structural change primarily. And I'll explain briefly what that means later on. Now, what happens when we get

the new technology and we get the structural change taking place, you're going to get some market forces in an open market economy that will push the economy back to a new equilibrium. The new equilibrium will have different types of jobs.

The simplest form of this is what we call in economics the solo growth model, where there's only one type of job and everything is just pushed into a new growth path of the economy. Here I'm more interested in the case where there are many different sectors of the economy and there is a sectoral relocation of jobs and the new equilibrium is reached.

The faster the new equilibrium is reached, the better off we are in a way because we're going to get there. But sometimes the speed of adjustment is itself

very costly activity so you have to balance out the speed of adjustment with the time that it takes you to get there. I mean it's an aside that you don't see, no not necessarily actually, the final bullet point is it says adjustment is usually very slow. Usually people are very surprised by how slow it is despite evidence that is in front of us and everyone knows. You know for example see

how long is it taking the countries of Eastern Europe that they were formerly planned economies, they opened up their markets in 1990, they joined the European Union

10, 15 years later, it's now 35 years and they still haven't adjusted. If you look at the countries of the European Union, all the countries down at the bottom are virtually all of them the former planned economies and they're still adjusting their growth rates

And their adjustment rates are higher than what we see elsewhere. So we are talking about very very slow adjustments, you know, they might need another 15-20 years to catch up with the rest of the European Union. Now, why is the adjustment slow? Well, this is where I get back to my work. I can't talk about anything in chronics without mentioning frictions. And the adjustment is slow because of frictions and there are three in particular that

We identified both in theory and practice. There is information that we don't have good enough information in labor markets usually to lead to fast adjustment. Therefore, we're holding everything back trying to find out what's going on before we make long-term decisions. There is location where are the new jobs being created and do we need to move?

And there are skills. What skills do the new jobs need to adapt to AI and do workers have them? I've recently been reading a bit of economic history of the early industrialization. More evidence will emerge later on. And I'm quite surprised actually how these do get mentions. If you know what you're looking for, you will find them mentioned in things that...

people said before, especially in the 1920s, just before the Great Depression,

There was a Secretary of State for Labour in the United States called James Davies, a Welshman, in fact, who moved from Wales to the United States and became Secretary of State. And he pointed exactly these things that he was saying, how hard it is for Labour to adapt to the new world of industry because of these frictions and all that.

The poor man used to spend his name, Davis, D-A-V-I-E-S, and the Americans forced him, if he was going to register and get an American passport, to drop the E and make it D-A-V-I-E-S, which he never did in personal correspondence, by the way. Only on his passport, because he really wanted that job as Secretary of State for Labor, served under three presidents. Now, what's the relative importance of...

AI, of these frictions for AI. To answer that question, there is something that might surprise you, but I'm very keen on this distinction here. I don't know if I'm the only person actually that makes it in the profession, but I really think it's important, which is that AI and all the other technologies are having two effects on jobs. The one is the one that replaces workers directly.

I have here as an example the paralegals write reports for the senior lawyers in their offices and now we hear of more and more lawyers using chat GPT to write their reports. It makes sense, obviously, rather than have young lawyers spending hours and hours in archive rooms. That's number one. Number two is that relative sectoral outputs change

because of the technology. The technology affects productivity differently across the economy. And jobs must get reallocated according to the economy's demands because ultimately the supply of goods in the economy will adapt to the demand or there will be an inequality eventually. And that reallocation always brings adjustment as well. In other words, people...

people leaving sectors that are benefited most by AI or not entering them as young people, as new entrants to the labor force, not because the jobs that they were doing were taken over by AI, but simply because there is no more demand for the things, the output that comes out of those sectors, and there is a lot of demand in other sectors of the economy. Empirically, it's very tricky to distinguish between these two.

But you cannot always make claims. Now, the first effect is one that obviously attracted most attention both in public discussions and in the economics literature. How many jobs will be taken over? Is my job at risk? You know, that kind of question that people always raise.

It's the one that scares workers, that workers fear most about new technologies and you could call it the Luddite effect because of the famous Luddites of the first industrial revolution and the ones that lost their jobs because of substitutions. The second one however is due to productivity growth as William Bommel in the 1960s, 1967 and more subsequently Rachel and Guy of VLC and I

I have written a paper to show that if productivity growth rates across sectors of the economy are different, then you're going to get a movement of workers from the sectors that are growing fast to the sectors that are growing slower. And that's why most sectors of the economy that you see today are in health and care, hospitality,

and other service sectors that are not experiencing productivity growth. You could call that the "BOMO effect" because he was the first person to point it out formally within economics. Now, there are large disagreements about the impact of AI on jobs and society at large.

large disagreements about these two effects. Hardly anyone talked, in fact, about the second effect, although I think eventually it's going to be the dominant effect. Usually, technologies are very negative. You may have seen quotes from Jeff Hinton, recent AI Nobel Prize winner, who thinks that within 10 years we may not be there. But given his age, he expects to go before the 10 years, so he's not too concerned.

The poor man is only 77. I don't know how he concluded that, but anyway. Russell Berkeley has been very negative. Lots of them talk negatively about the impact that technology is having on the economy. Now, the economies are divided about that. I'm very Keynesian, as you can see.

They keep saying in the long run we're all going to be dead. Well, we're going to be dead anyway, so what kills us? Why does it matter what kills us? But I think actually for the economists, despite-- maybe because I'm a Keynesian, I'm very optimistic as well. I'm saying, well, whatever happens in the long run with AI, the best we can do now is to embrace it and benefit from it and prosper, improve the quality of our jobs, improve productivity. Before, it's too late. Once it comes, let it come.

So to know what to do, to adjust faster, we need to identify how those frictions are affecting AI. So what's the biggest problem?

Well, we don't have much time, so I can tell you a skill. We have to pay attention to skills, and that's why this session is called skills. But information and location are also important, and I'm going to give you examples now that you see how important information and location frictions are, and then from there I will lead on to skills, and then I'll pass on to Mary who will talk more specifically about skills in Britain.

Now, we need to reskill and upskilling is what we need mostly. We have to learn new, more difficult things. This phrase that you hear so often from children, "I don't do maths," should never be heard again if we're going to adapt to AI. We also have to learn-- so that's upskilling, we're going to use AI. There's also parallel skilling. When we move into the sectors that will attract many workers,

because there will be demand for the output, hospitality and health and care are the primary sectors. And how you train workers to join health and care and how much you pay them given that the government has to find the money is a different question that luckily we don't have to address tonight because we don't know the answer. Okay, so here it is. This shows you how important

information is if you talk to workers, and we did talk to many workers about their jobs and how do they feel and with the new technologies, how the employers introducing new technologies. When they ask them what can improve your conditions now that these new technologies are coming,

They mentioned better communication with managers and subordinates, more transparency about company policy, better social relations with colleagues so that they can talk more. And only then they talk about more time flexibility, including home working in the four day week. And then number five is more money will improve conditions.

Richard Layard, if you ever read, he's very keen on this ranking here, and it comes from his work actually. He's in his quarters. Now the first three are all about information. Workers are not well informed about technologies, and they're worried.

Now, obtaining information about the company's intentions and knowing how to use it is key for these workers. In our research at the Institute for the Future of Work, we learned that the best way to communicate internally is through informal contact, not by the boss calling the worker to go to talk to them. They hate that, in fact.

And workers prefer informality, you know, going to have coffee breaks with them or lunch or something. In fact, I was reading somewhere who, one of the top employers here,

the CBI and said, of course, you know, this is obvious and it's very easy to say, but you can't imagine how difficult it is for the bosses to go and informally meet their workers and have informal coffees and talk about the company. Well, that is apparently not that easy. Now, what about location? Location is also a problem because

those who are doing the research and the new startups, they should be concentrated in a small number of places. There are agglomerations, positive agglomerations of externalities to this kind of activity. And if you look at the successful, you know, the two successful countries, of course, in AI research, United States, by far the leading one, and China very clearly just behind,

All this technical research, all the digital skills producing the research, producing the AI, they're constantly in three places in both countries. California, Silicon Valley, Boston, Harvard, MIT, and Seattle are Microsoft in the United States, and Beijing, Shanghai, Hangzhou in China, where DPC are located in China.

the Ant Group and so on, and St. John where Hawaii is located and Tencent and these other companies. In Britain, as you're going to hear,

shortly, where all this concentration is entangling what is known as the Golden Triangle of London, Oxford, Cambridge. If you look at Europe, it's scattered all over the place. And I think that's the main reason for the failure of Europe to keep up in touch with the US and China in this.

Otherwise, they invest as much, but as I was saying before, the US is by far the biggest investor. The United States invests 60% of global venture capital investments that goes into AI research, takes place in the United States in those three places. China, the second, invests only about 12% of global resources. And if you put the whole of Europe together, it comes to about 12% as well.

But there is no global, say, Stockholm. There is some research in London. There is some, but nothing like the big centers of the others that I mentioned. Now, the main friction is skill. In older industrial revolutions, this was not a problem. In fact, I don't think there was any industrial revolution where learning new skills was a problem. In fact, in the very first industrial revolution, workers had to downskill to take the jobs in factories because before, they used to be--

skills crafts people in the cottage industry. And when the Luddites were called, Mr. Luddite first and then his followers, were offered jobs in factories, less money than they were earning before because they didn't use the skills they had before, you know, making textiles and all that at home, then they rebelled. It was very unsuccessful rebellion, by the way.

I think it was probably uniquely a successful rebellion against machines early on because the British government supported them and not the machines, whereas every other government supported the machines, from what I was reading. When electricity came, people didn't have to learn anything new. Whether you press buttons to operate a steam engine or an electric machine, you don't need bigger skills. It's only now that we need them.

Now, what are the main skills that are needed in the world of AI? You're going to hear about that as well. There it is, first bullet point points that out. But generally they're related to understanding data dynamics, learning how to communicate the results, and the communication is very important.

And those leaving the tech sectors to work in other services, they will not be influenced greatly by digital skills. You're not going to hear very much about that this evening. You know, you go into nursing, hospitality, personal service provision and so on. It's still useful to have some basic knowledge because you might be given a tablet and called to manage your appointments on a tablet, for example. But that's only basic knowledge. You won't need to work with AI.

So what they need is empathy, not Python. Unfortunately it doesn't rhyme as well as I wanted it to, but I still put it there. Now, finally, I want to say a few things about, I think I still have five minutes actually. Yeah. About preparing for education, because these are a little bit controversial, but I really believe in them. Now language, maths, and other traditional subjects are still essential, obviously, for AI. There will be a shift to STEM subjects,

and broader education. By broader, I mean wider kind of thing, broader than the British A-level system for sure, which is pretty terrible in the world of AI, by the way. And schools and even... But I also think that schools and even universities should not specialize too much unless you really have the next...

the next Geoff Hinton, the optimist John Hinton coming out, in which case let me specialize, or the Demis Hassabis or something. In other words, unless you're really going to work in the research on AI, I think it's better not to specialize. The 100-year-old people that I mentioned before, including Davis and someone else whose name I forget, in fact, they emphasize that the education system should not be specializing too much.

And the reason, I have two reasons for that. One of the reasons they pointed out a hundred years ago as well, which is the first one, that technology is changing too fast.

and it moves in unknown directions, that's where the information comes in. So today's important specializations may not even be needed tomorrow. So don't put all the Rx in one place to specialize in some minutiae and then you will not be able to apply them because something else might come along. The second one is new in the world of AI, that AI is getting better and better in STEM tasks,

The technologies are saying that there is nothing AI can do now that is technical. And if you specialize too much in a particular STEM field, AI might take the job away from you and you'll be very sad, I guess. What you need to do in school, most of you are here, is learning how to learn in future life.

Get the background to learning. Acquire general knowledge in STEAM, by the way, that enables you to learn more on the job. A stands for arts. I don't know why, but I think economics is the most important subject that comes under A. Maybe we should put the STEAM double E instead of STEM.

The general knowledge includes the basic sciences and engineering and lots of maths, of course. Life learning becomes more important than ever. Successful companies put aside time for learning. Workers have to have incentives to learn. So how do you provide the incentive? Well, this is where my well-being comes in. If you provide good jobs for those workers, they're going to love their jobs, are going to love you if you meet them for coffee and tell them

By the way, when I talk about this in China, I say tea breaks, by the way, instead of coffee breaks, which they appreciate, but I think it's coffee breaks that employees need to get the adrenaline running. If you have a good job, then you perform more and more, you help the company, and you

They like flexible working practices. They want them to be sympathetic about life outside work, striking balance between work and family. All these things that we discovered that determine the good jobs. So there it is. The biggest challenge we're facing is the transition. Frictions of information and location can eventually be overcome. Skills is a lot more difficult.

I pointed out about education, but we don't have time to go into well-being, maybe next time. But I think, not I think, I'm convinced actually given the work we've done so far that if you provide jobs that improve the well-being of the workforce and at the same time you let them own the training, you can put it online or you can offer them alternatives, then they are much more likely to offer that and

and the company will be able to implement the new technology faster and the workers taking advantage of it and productivity rising faster. Thank you. - Okay, well I'd like to thank the LSE and the CEP for inviting me to give a talk in the presence of such an eminent person as Chris.

I have quite a different presentation. So I'm associated with the Productivity Institute, which is an institute, it's a network run out of the University of Manchester, involves lots of different universities. I'm also involved with the Economic Statistics Centre of Excellence. In fact, most of my work has to do with measurement. And what I want to do here is just to give you some numbers, some feel for...

differences in skills across... sorry, I got it the wrong way. Results from a recent project, I would say very much work in progress. And what we're trying to do is we're trying to understand the skill demands and supplies across the UK. So it's quite a regional focus.

And then I want to talk a little bit about skill provision through the education system and then about training and what can firms do to deal with digital skill shortages and barriers.

Now, what I'm going to do here is I'm looking specifically at digital skills. I absolutely agree with Chris that you need more than just these very narrow digital skills. You need communication skills and many other skills. But in terms of the kind of measurement exercise I do, it's very difficult to measure the more what we call softer skills. It's much easier to measure the harder skills.

Okay, so I'm going to show you some numbers, some graphs. So we call these digital technical skills because the word digital skills often includes the software type skills and AI and data analytics but also the soft skills as well. So we do focus on these technical skills.

I've just shown we have a division. What we do is we divide the skills into three groups. What we call developer, which are the more advanced AI, data analytics, et cetera. User skills, which is kind of digital skills that businesses need

to use software such as Salesforce off the shelf. So they do need people who are able to work with these software as well as people who

you know, have the ability to use AI. And then we have these basic skills and it is very surprising how many job adverts, advertisements ask for Excel and no other, no other digital skills. They just want people with Excel. It's still basic digital skills are still very much in demand. But in terms of the division between the three of these, the developer skills are probably are the biggest. Now I hope this comes out well and the, yeah, so, so basically this shows,

Our unit of analysis, geographic unit, is what we call travel to work areas, which is a measure of local labour markets. And you can see here, this is basic user and developer.

Basic and user are spread out across the country. So firms all across the country have very high demand for those two types of skills, but the developer is very concentrated in what we call the golden triangle area, which what the golden triangle is kind of defined seems to vary a lot according to the person. But it's generally London, Oxford, Cambridge, and around that area.

What we do also find is that these two types of skills, the basic and the user, have been growing quite steadily for some time. So there's large growth in those. Developer is a bit different. It's growing in London and the South East but not growing in many other places. And that's kind of interesting. So there is a concentration of the more advanced skills in London and that's where they're growing.

Now, another interesting thing that we looked at in this data set was we divided a first sample of our job adverts where we can actually identify what the qualification level is in the job advert. We identified graduates and non-graduates for about 40% of ads.

And what you see here is that the, and this is for developer skills, so a very high concentration, again, developer skills in the Golden Triangle, but in the rest of the country, for non-graduates with very similar skills, it's much more spread out. And so this is the idea that firms...

A lot of the types of firms that need these very high skills are located in this golden triangle, but firms across the country also need these skills and often what they'll do is they'll use non-graduates instead of graduates to fill those skills. And then just some numbers here, again, based on about 40% of the ads,

looking at the earnings of people with various types of skills and the red are the non-graduates. And the non-graduates, there's a big premium to graduates over non-graduates but that premium has been declining over time. And in particular, non-graduates with those very kind of basic digital skills are very much in demand. So if I'm to summarise this, I would say there's a high demand for all three types of digital technical skills.

but the developer skills are much more concentrated in this Golden Triangle area. But areas outside the Golden Triangle, they do have a demand for these skills, and what they seem to be doing is to be more willing to hire non-graduates to take up those positions to fill their needs. I have done a calculation just recently

to take account of the fact that the industry composition varies a lot across the countries, but that doesn't really explain very much of these differences across the regions. And then, as I said, earnings of non-graduates appear to be gaining on graduates over time.

Right, so that was the demand side. And then what we're trying to do in this project is to match that to the supply of skills. And we do that using, some of you may be aware of this, the longitudinal education outcomes, LEO database, which follows individuals as they progress through their schooling

into the labour market or remain in education. And this is a database which matches up data from the Department for Education for HMRC, DWP, and the HESA data for higher education. So we calculate, for any one year, we calculate the proportion of those graduating with STEM or digital skills.

And we do this at three different levels, at schooling, further education and university. And then the kind of skills we're looking at here, the kind of things that you can measure well are maths and logical skills, computing, programming skills, science and engineering and applied digital skills.

So we're looking at people coming out of school, further education and higher education with these skills, and then we're also using exam results. So we're not just looking at those people, everybody who does these subjects, we're looking at those who do them well, get to a level of A star to B in A levels or a 2.1 degree. And then what we do is we map that across the country. Now here we're restricted to England because the LEO database that we have access to

It doesn't include Wales or Scotland. And then you get, again, the highest concentration in the Golden Triangle. Now you might wonder what this place is here. You might have seen this in the earlier maps. So that's Leamington's Farn. And that is, I've only discovered about a week ago, it's the centre of the games producing industry in the UK.

So that's the highest in our chart. And of course, it's very close to Warwick University, isn't it? They get a lot of graduates from there. But we have, again, a pattern of concentration of the supply of skills within the Golden Triangle. But you do get some other areas where there's a lower concentration, but still above average in the West Midlands and Manchester, and Birmingham in particular.

But you have large areas of the North East, Yorkshire and East Midlands with below average supply of digital skills. So we now have a demand and a supply, you know, a kind of mapping of both of those and then what we do is we put them together. Now this is very preliminary work, I'm not going to concentrate too much on this but, so I've just divided these into four quadrants, high supply, high demand.

These are the Golden Triangle, plus a bit more. Places like Manchester now and Birmingham are in there, and York.

high supply, low demand are mainly rural areas, so not so interested in those. And then the orange ones are the most interesting because they're areas where there's a high demand by firms for these digital skills but the supply is low. And these are often places where it's difficult to attract graduates into those places for lots of reasons, maybe wages, maybe amenities, but

they have to rely very much on people within the area to provide those skills and they generally have a low supply. So these would be the areas in this kind of mapping that you would say where the skill shortages really bite because they have a low supply but the firms are there and they are demanding skills.

And then the grey are those areas, or we often call them maybe the left behind areas, or areas where skills are a problem. There's a low supply, but there's also low demand. These areas have lots of different issues and sorting out the skill problem is important, but it's not going to solve their problems. However, some of these other areas, it's quite obvious skill is a particular issue for these areas.

So that's just giving you an overview of the kind of work we're doing in trying to understand this geographic distribution of demand and supply of skills. And for the rest of my talk, I just want to focus on the provision of skills and how do firms get the skills they need, given that there are skill shortages. So basically,

There's two different ways. One is to the education and provision system and the other is to the training, firms provide the training. So I'd like to emphasize here and I think Chris also emphasized this, schooling is very important. You have to have

good STEM skills, good literacy skills in schools. Otherwise, not just in schools, but also in FE colleges.

If you don't have that if an area doesn't have those foundations if they don't have people getting to a certain level in schooling then how can they ever really attract the firms and Grow increase their productivity and grow so

There are many areas of England where the proportion of schools achieving high grades in these particular subjects, Math, Science and ICT, are well below the national average. And these areas kind of map on well to some of the areas I showed earlier, some of the grey but also some of the orange areas. So I picked out a few places here. The share of these pupils with these skills

in places like Hull, Lincoln, Liverpool, Sheffield, they're about 75% of the London level, which is low. So there was this huge variation across the country. And this, to my mind, is an important area for public policy. And then...

There's a lot of evidence about the importance of digital skills even below the degree level and this puts forward the importance of further education colleges and I don't need to, there's a huge issue to her in this, I know Sandra is here in the audience and she's contributed a lot more than I have to this but these colleges now seem to have an increasing role in this digital world and there are places where it's well known they're underfunded

And the whole education system is very fragmented. So the Skills England recent report on this really highlighted the need for joining up education across these various levels to make sure we have the skills that we need. There's an education problem here, an education provision problem, but

Changing education provision takes a long time. So in the meantime, firms have all these demands for digital skills. So what do firms do? Well, one thing they can do is to provide training themselves. I've just put a couple of figures on here from the Employer Skills Survey. The first is expenditure on training by firms. The second is average number of days per trainee.

And this goes from 2011, I took these numbers very quickly from the ESS, I probably could have done something a bit better. But I looked at this many years ago, I have a paper, and this downward trend is going on a long time. It predates the financial crisis, I think to the early 2000s. The firms seem to be training less. So they have these demands for skills and they seem to be training less. So this seems quite pessimistic, but

I think there's a recent trend that's probably brought, you know, accelerated by COVID, that there's a positive trend, and this is online training. So...

The ESS has one statistic on that. It says 67% of training provided to employers had arranged some online training, and that compared to 51% in 2017. There's a very good CIPD report on this, and they give some more other numbers as well. They suggest that there's a lot more firms using online training now than there was pre-COVID years ago.

Most of this online training is IT and digital skills, but there's still quite a lot of online training in the communication skills, leadership skills, and all the things that go with the IT. And I've given you some logos here. Some of these are...

AutoCAD and Salesforce specific to the software that's being produced by these companies. But there's a couple of companies... This whole market seems to be concentrating quite a lot and very rapidly. So LinkedIn is very involved in this and Skillshare is another one and Udemy, which I'd only heard about very recently myself, but it seems to be one of the big providers. So this is one route for firms to take and I think this is...

What's positive about this is that there's a well-known argument that firms will underinvest in more general skills training because they can invest in training for the worker, the worker leaves and goes to a competitor firm. So the more general the skills, the less the investment. And one of the things that online training has advantage over traditional training is that...

It lowers the costs of the training and it does it in two ways. One is cheaper courses. Now as these firms get bigger and concentrate maybe the difference in price between the courses, the cost of the courses to the firms probably is not so large. But when you look at what is the cost, the other part of the cost to firms is the opportunity cost of the time a person spends away from the workplace getting training.

We do a lot of work on measuring these kind of skills. If you look at the two parts of that, it's the opportunity cost that is actually the larger cost to the firms, and particularly small firms because they can't do without the workers. They need them to be there. So what online gives you is that

you can reduce the opportunity cost by quite a lot because workers do the training in their own time often or they'll do it in a slack period. And so the opportunity cost goes way down. The firms and the workers in some sense are sharing the cost of the training by the workers using their time and that should lead to much more training. Now there's not a lot of hard evidence on this yet. It may take some time for this to emerge but

There are some barriers, digital, the competence of instructors and lack of information on the quality of providers, especially for small firms. So there's huge numbers of people out there providing online training and some are better than others. Information barriers can be a problem, but information barriers are something the government can actually do quite a bit, our policy can do quite a bit about it.

Also, the CIPD report does make this point that learners with weak academic backgrounds or from low-income backgrounds tend to gain less from online training. So it's not a way really of bringing more equality, but it is a way of getting more digital skills. Another way firms can use digital...

can gain digital skills is to use digital platforms. And they've been around for quite a long time, and that's a cost-effective way of getting the talent you need. So about 50% of firms outsource some IT services, and this has been growing very rapidly. This is part of the gig economy, but it's a different gig economy from the Uber driver or other types of jobs on platforms, because a lot of the people on there are young people who are gaining a lot of experience by

doing this work on the platforms and then getting full-time jobs afterwards. So it is, but it's been there for some time. A relatively new

source of skills for firms is crowdsourcing. There's very little hard evidence in this because it's very, it's new. It's obtaining work information or opinions from a large group of people on the internet. So you're not necessarily paying for that, you're just going on the internet and getting information. We have done a little bit of work on TPI on this and we find that, it's a small scale survey, but it does seem to be for small firms

This crowdsourcing does seem to be an important way of getting information that kind of is very far from their capabilities. So they know there's new technology out there and they use crowdsourcing to find out some more information and then if they decide to adopt, they maybe either use digital platforms or other ways of getting their skilled workers. Again, as with online training,

digital platforms and crowdsourcing have this problem that you need information on the quality of what's being provided and that's something that small firms really do struggle with. Now I just want to end with this and this is something I often present when I'm doing presentations and lots of different things because it's something that I'm starting to work a little bit on. It's something that social psychologists really do emphasise a lot. So digital skills are often

It's often young people who have the most advanced, the most up-to-date digital skills. So what happens if there's a problem with your young people? There's this social psychology literature that is beginning to get a lot of traction. Now, you may have heard of Jonathan Hyatt. The Coddling of the American Mind was his most famous book. But now this one's The Anxious Generation. And Gene Twenge...

She's also quite well known, but she's much more quantitative. Both of them have the same kind of argument that digital technology has caused some change in the way, in the attitudes of young people and the way that they work. And particularly Generation Z, they focus on Generation Z. But these generations, they're not so set in stone. But people who've always used digital technology

grown up with a smartphone, always on social media. And Jonathan Hyatt, not so much Gene Twenge, also focuses on the overprotective parents of which I would consider myself one. But these works suggest that

there does seem to be a problem or an issue with young people having much more mental health problems than previous generations. And so if you're young people, if there's an issue, then there does need to be something done to address this. Our own work suggests that Gen Z are much more likely to value leisure time. That's maybe a good thing. But we have, you know, one of the biggest problems

most important things that is going to happen over the next 30 years is the ageing society. This is a time bomb and if your young people are not so engaged in work, that's a problem and as I said, they're the ones with the digital.

skills. So, just a conclusion, there's a large geographic dispersion across the UK. There's evidence of a move away from graduates to those skilled through alternative means. And I should here mention that yesterday the OECD published a really interesting report on this called Empowering the Workforce. And they've really picked up on this idea that if you want good jobs, having a degree from an LSE is becoming less enough and

What you have to do is you have to do all these online training courses and get your CV up. So I would highly recommend that. I couldn't put that on the slides because it only came out yesterday. Education providers need to adapt and join up. But firms are adapting. Firms will always find some way to adapt, and they are adapting through online training. But there are some worrying trends for young people. So that's me finished. APPLAUSE

Okay, great, thanks. Both speakers were absolutely brilliant in terms of delivering their time, 25 minutes each. Okay, so I think we can open up to some questions. A lot of show hands. I think probably it's best if we take two or three in each go. Maybe we'll take two or three from the audience here. Martin's got some online. We'll take...

I've got a couple there as well. So why don't we just begin? Have we got the microphones yet? Okay, let's stop. We've got the nearest. It's not going to be two of them. It's two. Chris wants two. We could maybe get chat GDB to do something to make it. Okay, we'll do two at a time. 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. Okay, go.

Thank you so much for this amazing discussion panel. I had so many questions with every point. The presentation was brilliant. You're only allowed two. But I'll keep it limited to one. So my name is Shryansh. I studied development economics here at LSE last year. And my question is very limited to the education part of things.

So now many people sort of believe that with AI, the most logical part of doing an activity can be very easily augmented to AI, which is related to math, stats, anything related to logic and reasoning.

And in that sense, how important, what are the implications of a child learning these basic logic reasoning and things where in a world they are seeing that these are the processes which can be very easily augmented to AI? Then how does a child need to handle this part of the education? And how does he or she need to deal with this changing environment, basically? I think this sound went up fast. We'll take this as a second, then we'll take some responses. And I may engage as well.

Thank you for having me and taking my question. I'm a senior at a high school in Texas. Both of you had large thoughts about education, considering A, this generation being the anxious generation, and B, the fact that we have to counter inequality caused by neoliberalism, which last week's lecture by Professor Asimov, who spoke about how do we counter both inequality

Can you hold the microphone? Sorry. How do we counter both at the educational level by giving skills that will both deal with previous inequality and future ones, as mentioned by you, due to low income and lower scores to access online skills training? OK, good. You want to go first? I can answer that quickly, actually. Because that sounds like a good question, but in fact, it's not very good.

Because the skills you mentioned, I mean, it would not be a good thing to do, let me put it that way. Because the things that you mentioned, you need them to learn other things. You know, when I said the best kind of education is to learn how to learn in later life. If you go to work for an employer who needs digital skills, and then the employer tells you,

you know, here is what you are going to learn online and here is what we're going to apply and all that. That would be built on engineering, physics, maths, and all that. And you're going to say, oh no, we didn't do these basic skills because they can be done by AI anyway. Then you've removed the foundation of what you're going to build on. So it's essential that you do it, but don't think that that's all you're going to learn and get a job just because, well, you might get a job just because of that, but you're not practicing it.

I mean I just agree that these are foundational skills. AI can do so much but it depends on utility sometimes. AI makes a lot of mistakes, now maybe that'll get better but I mean any of you who've used it, tried to summarise stuff, seen how they make up references and all sorts, you need the skills in order to check.

People talk about coders are going to be redundant, but you need coders because AI can give you the code, but if you don't understand the basics of coding, how do you know what AI is giving you? All of these skills, they're very important. It's what Chris was saying, you build on those skills. We need more rather than replacing all our skills are redundant. Did you get the question? The question here was more about inequality dimensions of

as I understood it, I think. Is that what you wanted to ask about? And then we've referenced... Yeah. Well, inequality is going to be a big problem. It's going to be a bigger problem than we've had so far. And unfortunately, we don't know how to deal with inequality in most countries because of the...

redistribution which is one way through social services the way that Swedes do it for example is something that

It's a non-starter because it's going to lose you both. Can you imagine Trump saying, make America great by redistributing, by reducing inequality, putting rich people down. So it's a serious problem. I do say to students actually when they say, well, can we specialize in labor economics these days? I tell them, inequality might be the only good way we know is the Swedish way where you distribute...

You don't give money to those on lower incomes, but you provide good quality services like childcare services to encourage more participation, education and so on. But for that to happen, you have to have full trust in the government that will provide you good quality services. And unfortunately, not many governments have that kind of trust that they will actually supply services and that will not

pay more attention to lower taxes than spending money on good quality services. We can be a Labour government. Which we all love. It's so different.

Thank you very much. That was as entertaining as I remembered Professor Piscirides. I used to sit here in your class ages ago. So I did your economics class ages ago here. So thank you. This was as entertaining. You still remember you well. Yes. Parts of it. But my question is more around the later stages, the job creation. So my colleague and I are building an AI-first skills-based job matching platform.

And it's really, really hard. It's the best of times to build it. It's also the hardest of times to build something like that. People don't really understand what skills will be available. People don't know what skills to hire for. And I think there is a bit of a knee-jerk reaction in terms of how do we cope with AI? So what would your advice be to employees and firms who are looking to hire

but they don't really understand all of the contexts of digital skills, but equally have never really thought about the unit of analysis of soft skills. And I think that would become more important because if you have AI co-pilots working with us all the time, which we do, what is that proportion of soft skills versus hard skills that an employer should be looking at?

Thank you. Hello, thank you so much for the great presentations. We heard a lot about upskilling, are we skilling?

But in a recent OECD, Adult Labour Markets Skills Survey, one of the kind of highlighted conclusions was that the UK, according to that report, has an over-qualification problem. I know some of your colleagues have critiqued this report, but I think it kind of made me think, are there any particular skills which might have been in demand in the past that are going to be considered less useful on over-qualification in the age of AI?

Now, how should individuals that might otherwise try to acquire those skills or the education providers that taught those skills in the past adapt to this?

Let me start with this one. You do have skill shortages and over qualification happening at the same time and the over qualification tends to be also concentrated in the same places, the high demand for the digital skills.

It's generally to do with labour mobility and in particular graduates moving to places like London where there's a demand for one kind of skill. They don't necessarily have those skills and they often end up taking non-graduate jobs. So what do you do about that?

That's too difficult a question to answer because people go to university for all sorts of different reasons and we don't want to say we should have no history departments or whatever. I think this whole thing of online training, in the OECD report that came out yesterday they really emphasised this, that younger people going on the job market

need the qualifications, need the degrees from the good universities. They need more than that. And employers are kind of expecting more than that. So in some ways you have to invest in different skills and different parts of your education pathway. Which I think comes to your question as well because, I mean, neither myself nor Chris could really answer what proportion should be soft and hard skills. It actually needs to be both.

And if one part of your education pathway has given you one part of skills, then in some ways it is up to the individuals to get the other part. And sometimes the firm will pay. So they'll be lucky enough to get in a job in a firm that's willing to pay and pay for the accreditation. Yeah, I'm not... You shouldn't be too worried, actually, about...

skills that will become obsolete. I mean, skills are becoming obsolete all the time. We all have obsolete skills. You know, if I look back in my career, I have ever so many obsolete skills that I learned and did and then I relearned. But so what? You know, you move on. Always look forward. The main thing is what I said before, that the

when I was presenting that you should be prepared to learn new skills all the time. And it's unfortunate that not many workers have the opportunity to learn on the job because that's the best way to learn about the company. I mean, we are very lucky because, us working in academia, because we can learn all these new skills on our own. We can take as much time as we wanted. And in fact, I said to you, I've been working on the well-being from work. I mean, you won't be surprised if you see...

happiness happiness surveys when they ask people the jobs and how happy they are about you know job satisfaction i mean who scores highest in terms of enjoying their jobs and job satisfaction is academia and i think it's because you know these professors and i think it's because they can do things on their own and take initiatives so i shouldn't worry about that about that just just don't specialize too much too early on this oh the other the other thing actually that i believe very strongly in that

applies to both is that I think you should be doing things that you are enjoying doing. I mentioned before that if you enjoy it, you feel better and you move on. So like you when you say your friends, what should they look for? They should look for people who are prepared to learn. Obviously people, if you can identify intelligence in the intelligence and they should get people who are genuinely enthusiastic, not making it up.

who really love doing what they're going to do. And then you know you're going to succeed if you go like that and move on. Yes, this is right. So of course, chap GPT could be writing Pissarini's matching book.

No, no. Actually, if you try it now, we'll write it. I tried it. Okay, enough of that. Let's go online on the basis of that. But it also tells you where they got it from, right? Oh, yeah. Okay, so questions from online for you now. Vandad Pobarami over in Canada asks, how do you, Chris and Mary, use technology and AI tools yourselves?

There's of course a lot of interest in the large language models and the way that they like to predict what people want to hear. How do we ensure that people beyond the labour market have skills in terms of critical and social skills to ensure that they're able to see how to react to that appropriately?

Say that again, the first question? I thought it was one question. First one was, sorry, there were two there. How do you yourselves use AI? Or how do we use AI? Well, we use it all the time, actually. Especially large language models. I mean, I use it and then I check them out.

I'll give you an example. I'll give you the most recent use of AI that I had. I got a document to sign and it was in Swedish. Of course I didn't understand anything. So I asked the GPT to translate it for me into English. The first time he said, sorry, I'm unable to translate this document. The second time he said, your document is corrupted. It's got lots of symbols that are like this and then produce all those...

And then I asked Deep Sik to translate it and literally within 30 seconds I got the perfect English translation in simple English. Everything. So there you go. I read it, I agreed, I signed it, sent it off. It tells you something about where China is going. They copied Judge EBT and then they leave from...

I have to admit that I'm not very good with technology. So when I have technological problems, I ask my daughter. I put up with her raising her eyes and thinking how she is so incompetent. And my research assistants of which I have many. But like many of us, we've used AI to summarize literature and stuff. I'm not as impressed as most of my colleagues on this. I've seen good and I've seen terrible.

I have colleagues who just love this. You have a second? Was there a second question? You go with it. It was about ensuring people are able to maintain critical skills rather than labour skills in the face of technology. I want to refer to that actually. I think those skills have undervalues of empathy, as I was saying, that is absolutely necessary. We're putting...

too much emphasis now on the technical skills at school. I mean, traditionally, we never paid much attention to those skills anyway. I mean, I know there are some schools, because I've been trying to follow what they do. There are schools that do charity work and they help people with special needs and they develop these skills. But I think those skills will be in demand increasingly because

As we become wealthier, if you like, as AI is making us wealthier, we're going to spend more and more on these two big sectors, on health and care, and of course ageing contributes to that as well, and on hospitality. The demand, to put it in economic terms, the income elasticity of demand for both is bigger than one. So GDP going up by...

and this will go up by 1.2% usually for health and hospitality even more. And that means we're going to need more workers because those cannot be mechanized. I mean, I have talked to economists that I respect and they're saying, oh, they're humanoid robots and they're going to do all these jobs and all that. But I just cannot see. So we need to learn those schools. And it's up to the schools to

So we find that when we do our surveys of firms and talk about adoption of new technology and such,

You often find that there are people within a firm who are very enthusiastic and they have great technical skills and they understand the technology, but they don't really communicate that very much to the chief financial officer or the CEO, and they're the people who are making the decisions. So it seems like the more complex the technology, the more you need to be able to communicate what it's all about to other people, and that also, I think, is an aspect of this where these kind of skills are very, very important.

Good. Let's go back to the audience. Questions? Sandra can ask a question. Hi, thank you. I just wanted to ask, what do you think that AI, generative AI in particular, how does it affect how you think we should teach at university? The courses that we all teach and the ways that we assess students, particularly with things like essays where chat GPT is an excellent job in many of my undergraduate types of assignment. There's one here as well.

Hi, so I'm not an economist, I'm a designer. And I think most people kind of agree that human creativity is by far and away still the best thing. So the cutting edge of human creativity is brilliant. But as a designer, the journey of a designer in a career sense is it sort of starts as a hard skill, but the better you become, it becomes a more human soft skill, more intuitive. So in terms of a

The journey of a career that's sort of kind of skills-based like that, it's a long-term investment for a firm, which can be very expensive, quite a long waiting game that might not pay off, when AI can do the sort of grunt work that you might need to do to sort of cut your teeth and learn how to do the thing which ultimately becomes the valuable thing long-term. Does that make sense? Which one do you want, Mary? Or both? The designer, or what we should do about AI in teaching and university?

Do we need to teach AI in universities? Don't the students pick these things up themselves? Oh, so you think AI is going to replace us? Well, no, no, no. I mean, it's very hard to know sometimes how you adapt your teaching and assessment because AI is very good and you can't just... If you don't really need to remember basic knowledge and that's not what you should be teaching students, well...

How do you approach that? I mean, this is an issue that I face in my own classes. I mean, how do you with them? Well, particularly because I have to do a lot of exams. I really don't have a good answer for that one. I know there's lots of people in my university and presumably here too trying to work through that question. I keep getting emails about how I can go on courses and how to use AI in my teaching, but I don't teach anymore, so I ignore them.

Chris, do you want to do the use of AI in design? And I think the implications about improving long-run productivity from using AI to start with. Yes, sort of. You see, both, I think in a way they are related because you said how much AI. I think we should be using as much AI as we could in teaching as well as in design because it will improve productivity. But...

You know, I don't think... It's difficult to tell with creativity, you know, by avoiding to sound like a snob kind of thing, because however hard AI tries or computer graphics tries, it's not going to be as good as the one produced by human ingenuity.

We're never going to see a Leonardo da Vinci. That's my point, is that for every Leonardo da Vinci, we're saying that he's good creative, but there's a massive amount of ad creative. Yeah, it's what I was coming to, but the things you see hanging in offices and houses and all that, I don't think you could tell the difference. And it's going to come. Yeah, I mean, that...

I mean, it's going to take a lot of jobs away of the kind of artists that work in Camden markets, which are flourishing up to now, I guess. But you said something about long-term investment by firms, though, which I'm very interested in.

long-term investment short-term investment in different sectors, in different occupations? What was that? What I mean is, in design, student designers are very enthusiastic, but they are like crap. So it takes a while. They need to be working on live briefs, not necessarily doing the really expensive stuff, but doing some of the grunt work and

kind of fiddling around on things, which is already being done by sort of AI iteration. So it takes a long time before you're trusted as a designer to get your intuition right. You know what I mean? It kind of takes a while for you to develop that taste as well as that skill. Well, yeah, but that's not due to the technology. Isn't that always a problem with that profession that you labour away on the breadline for 30 years? Yes.

until you get some reputation where you can put your prices up and have a living wage. Is it different from... You're copying other artists up to there. Now you might be copying products of AI, which I don't see any problem. No, it's not so much copying products of AI. It's that...

there will not be enough of that grunt work to do, to give off to junior designers and say, "Oh, go and work on this thing. We need 500 sort of logo iterations. Go and do that." AI can do that work, so there is less demand for that junior role. -You get what I mean? -No, you're right. That's the lawyer's problem, isn't it, that I pointed out? You don't need the paralegals now. And then how do you... You lost the stepping stone.

Yeah, I know that's a big problem with many professions actually, including service accountants. But I can tell Sandra about use. Sorry, that's switching now. But I agree with you. It's a problem, which I hope you don't have to face by the way. You know, it's always the case actually when new technology comes in teaching. In my first visit to the United States, which was some time ago, it was when...

cassettes were first invented in a big way for recording. And the person I was visiting, a very distinguished American professor, I'm sure you've heard of him, I don't know, maybe I shouldn't say his name. Depends what you're going to say. No, it's not bad. He said to me, this is the future of teaching. Someone will come here who knows about recording and lectures, let's both of us record our lectures on these little things.

Then take 10, 20 of these when you go back to England to the LSE, just put the lecture on, it would be shown on TVs around in different rooms remotely. You don't need to work anymore and you are going to make a lot more money if you are in first. Well, luckily I didn't do it, but he did it because it died before it was even, before it even reached sort of adolescence.

Then when Wikipedia came along, it was when I had children at school, secondary school, and they were giving them homework to come home, and the teacher was saying, "Don't go and copy from Wikipedia now, you have to do this from your books." And then you think, why?

When they realized that it was copied, because two of them had the same phrases that they got from Wikipedia, they had marks deducted. Well now, they tell them one of the sources you can use to write your essays at home or your homework is Wikipedia. They're online sources. They all walk around with their laptop. So the same will happen with AI. I think it should be used. The remarkable thing is how the

The popularity of teaching sort of person to groups of person to person hasn't declined. If I put you a recording from YouTube from a previous lecture on these skills here, would you have come and watched it like that? OK. You had a question before, didn't you? No. Yeah. Yeah, you see, I remember. You got too tired holding your hand.

No. And then next door to you, this is the

Hi, thank you so much. It was a lovely talk to hear. I just had one question mainly on your research that you had presented about digital skills and the fact that if more digital skills are being learned and by nature more people want remote work, then shouldn't the shortage be addressed across geographies where someone in London can do work for something in Scotland where the demand is there but the supply is not?

So what is actually affecting those barriers when with more people wanting remote work and more kind of digital skills coming through, what kind of prevents that barrier of skills? Am I understanding it wrong in terms of education or like more on the workforce? And what this might mean in terms of pay and incentives as well? Yeah, so I mean, does it... Let's take a second. Yeah, this is a question of mobility. Oh, do you want to take more questions? We'll take a second question right now. Sit down next to the guy who had the question.

You were the first.

Alright, thank you very much. I just wanted to ask because Professor Pissaridis touched about the location as a friction and then Professor Mahony spoke about distribution of skills. And I wanted to ask what should be the right approach when it comes to preparing the workforce for the digital transformation?

Because we have a workforce that currently represents five generations, it is a multi-generational workforce, more than ever before. So the question would be what businesses, policy makers, educational institutes can do in order to help people transition to digital skills or the skills that are equivalent for the future, given that. Okay, Bonnie, you go, maybe. I don't know.

This was more about automation. We couldn't hear the microphone very well. Even though you got the microphone, we couldn't hear it very well. Sorry, I can do it. If you can do it in one sentence. Yeah, say it in one sentence.

How do we navigate to a multigenerational workforce in order to gain the right skills for AI for the future? How do we motivate them? How we navigate as a whole. How do you navigate? I think it goes back to what you were saying in our presentations.

You wait and let the companies take initiative on the training because they know best what are the new skills that are needed and they also know what kind of machinery they're installing and what kind of skills that they're going to need. And it's best to do that but then you do have the problem of under-training if you leave it entirely on the company so there should be help from the government. And it's one of the things that Britain failed dismally actually over the ages to do.

- Especially the vocational side. - Yes, especially the vocational side. But that's what's needed. For as long as I can remember in this job and all that, we've always talked here, why can't we get more people trained? Why can't we find ways of helping companies do it more? And the first question was related to that. Well, the answer on mobility, it's a very big problem.

In an ideal world, what you need to have is to have concentration of R&D discovery because of the positive externalities. But then you have to have good ways of spreading, of diffusion across the country. And unfortunately, people tend to...

move more in the locations where the sort of trendier, more fashionable, more impactful work is taking place. Ignoring that the diffusion also requires skill and ingenuity to do and it's also kind of R&D. And it is a problem with mobility and also housing plays a big role there. If you

If you are in London and you are inside the housing market and you live in London, would you go to Scotland that you mentioned? I mean, not here, I guess, Scotland, but... I mean, it's difficult. People don't do it very much. I think, actually, the United States is probably more successful than other countries, certainly much more successful than China, Europe, the other economies, because...

The research is done in California, Seattle, and Boston, but Texas is very big on diffusion. Texan companies are doing a lot of applied research there. And those are long distances, but the Americans do move a lot more than anyone in Europe.

I think that, you know, people move because of wages, our real income, so housing has to be taken into account. But there's also this amenity argument and this kind of agglomeration of people. Young people seem to congregate towards the South East. But what's interesting, what's been happening is... Manchester, I think, is a very interesting case because there is a lot of movement now, people to Manchester. And somehow, how did Manchester...

get to that position. That's something we're kind of trying to grapple with. I mean, some people talk about and Birmingham is also showing kind of similar trends and going to the earlier question about education provision. One of the things

answers that people give is the two Andys, Andy Burnham and Andy Street. They were very important, you know, focused on the local community to get, you know, everything, lots of things need to come together to attract people, to attract the firms. And if you have strong

people in government like that maybe maybe that's what what helps and then you start to get people moving and then Manchester becomes the place everybody wants to go to and then you get you know kind of they feed on themselves so I don't think we understand this very well but I think we have now we've got some um we could see something happening in this country and not to contradict Chris but I'm I think that you know the US was very good at mobility

up to about 15 years ago, a decade ago, but there's a lot of evidence now that mobility has actually decreased in America. Yeah, but it's still well above what we have here, though. I mean, it's been on a declining trend for sure, but it's still high. Okay, a couple of quick last questions. One here. You want to ask him a green day shirt? He wants.

- Thank you for this intellectually rewarding presentation. I'm a current master student here at LSE studying dual degree. I just have one question to address to Professor Christopher. You mentioned multiple times during your presentation about the Luddite effect and I'm currently doing research on Luddism in the context of AI as well. I'm kind of curious from your perspective, what does Luddism represent or mean?

Our current age of AI technological change, would it be the same as before during the first industrial revolution when those workers stand up against machines or could it be a bit different like when some scholars would argue it's when workers standing against capitalists who use machines to oppress them. So just kind of curious on your thoughts about this. President.

My name is Aman. I'm from India and I actually have a question that's more adding on to other questions that have been asked because they've been asked. But basically in all these areas where you are saying that there is high supply and low demand, will demand increase or do you think people should start migrating so they don't end up in jobs below their skill level? Should they stay there? Will demand for technology increase?

Was the phrasing clear enough?

I mean, this is similar to the previous question. Why do you get some areas where people are... Why do you get mobility with some places, some eras, and not others? It's very difficult to understand. I think there's... If you look at my kind of four quadrants, the ones that are difficult, really difficult to understand, and the ones we're most worried about, is where there's low demand, so the firms don't want to go there.

And the education system is very bad.

The graduates don't want to go there. What do you do about places like that? I don't know whether that really helps. This is an age-old question, of course, about why employers want to go to certain places. I've studied it quite a lot. The question you mentioned, actually, it's a much bigger question. It's known as the machine question sometimes. The Luddites...

it was the most high-profile one. But as I was saying, the reason they were rebelling was that they were getting wage cuts. They were losing their wages, not necessarily losing their jobs, because their skills really did become obsolete, and they were mainly textile workers working at home, making the textiles from yarn or furniture makers. And the reason I say it's got a long history is that as we go...

As you go through economic history, there are periods where this was a really big question. And governments on occasion took completely hostile attitudes to this, especially early on. In fact, the British government of late 18th, early 19th century was the first one that was sympathetic to the machines because it wanted to be sympathetic to the new industrial class.

ignoring the workers completely and that's how the first industrial revolution came. Then another time that it was a big question was in the late 20s early 30s when Keynes referred to it in fact, he refers to the term of technological unemployment. It was the first time ever that you saw the term technological unemployment being described because

machinery was making workers unemployed, which is not necessarily true, but that was taken over by the Great Depression, or rather that question stopped completely at the beginning of the war. They thought, you know, that's it, I have no problems anymore.

But then the question was reborn with automation. The word automation was invented in around 1948-49. And then they started long discussions where automation was going to get rid of labor. In fact, John Kennedy, the president, was very active. Classic quotations. He took very, very strongly the side of the machines.

It's a famous statement, in fact, that I cited in one of the early papers with Mortensen that said that if men, he was referring to men, but that was this, in fact, if men have the ingenuity to invent machines to take jobs from workers, then surely they have the ingenuity to invent new jobs for those workers, which is the classic case that supposedly... So it's a big question. It's...

It's worth studying it. Are you doing it for your PhD? For my Masters. A Masters, huh? Are you prepared to read a 400-page book? I can recommend you one. Okay, okay. I'm going to test you. Okay, great. I'm afraid we've run out of time. We've actually gone over a little bit. And I know there's more hands. Maybe we'll get another opportunity. So I think we should thank the speakers for fascinating the session. Thank you.

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.