good morning everyone i'm andrea fracasso i'm a member of the editorial committee of the festival in trento i'm a professor here at the train to university i welcome you all today we carry on the debate that we started a couple days ago on the topics of this festival the title of today's uh meeting will robus take away human jobs is definitely one of the most interesting everybody is wondering whether technology besides changing jobs will ever have an impact on wages as well considering data available we can evaluate this impact on different tasks carried out by workers in different industries so on the grounds of such data we may answer the question whether technology or robots will ever replace human beings and through this analysis it is possible to understand whether rebels and technology will strengthen or increase employment whether they are going to really replace a certain jobs as we are going to hear in the today's contribution but i'm not going to tell you anything in advance the answer to this question varies according to time space and it also varies according to worker and tasks on the basis of the skills and the activities carried out by such workers the results of the analysis that you're going to hear about today are based on a very in-depth study because it's not just technology that affects jobs so those the um globalization of trade that can in the globalization production can have an impact on the labor market so today we are lucky because we have a professor dorne who is one of the experts in this field together with his co-authors was one of the first to develop this type of thinking and today is going to report on the results of his study and we'll be talking about the relationship between international trade technological innovation and labor so it's my pleasure to introduce professor dorner he's professor of international trade and labor markets at the university of zurich who is a really an expert and is also affiliated professor at the ubs international center of economics and society he has been visiting professionally in the most prestigious american universities like harvard university boston university mit and the university of chicago is the member of uh the editorial board of two important publications he is a researcher fellow of the centers of economic policy research in london of the institute for the study of labor in bonn and the center for economic studies the ife institute in munich professor dorner has always focused on the topic i mentioned namely the relationship between labor markets international trade technology and the impact of all of it on maker economics today's contribution will however be confined to other topics but a topic that professor dorne tackled recently is the effects that all such transformations have on the labor market as well as the impact on the outcome of political elections particularly in the united states i know he's not going to deal with this in his contribution but during the debate we may ask questions about these subjects so without further ado i would like to thank professor dorne for accepting our invitation i'd like to hand you over to him it will be walking around so please use the microphone ladies and gentlemen thank you for coming to this talk and andrea thank you for this very nice introduction work in the form of paid employment takes a defining role in our human life when we look at the distribution of income in the economy we see that about two-thirds of the overall income that the economy generates is going uh to workers income and since the labor income is more evenly distributed over people than his capital income that means that a very very large fraction of people are living primarily of labor income moreover labor is not just important as a source of money labor is important for people in the sense that most people spend a large amount of their time in the workplace it is a source of social contacts it is a source of people's feeling to be valued and as a consequence it is a concern for society if labor and jobs are being threatened and there is in the last several debates the number of trends that really make us somewhat worried about the prospects of workers we see in many countries that economic growth has become fairly slow here on the left side we see a chart from italy where over the last 10 years per capita econs have stagnated and while the overall pie is no longer growing much it is the case in many western countries that a shrinking slice of that pie is going to workers that's the chart on the right hand side which shows the fraction of italian income that goes to workers as opposed to capital moreover this shrinking slice of shrinking pie that is going to workers gets more unevenly distributed among the workers which means that a large fraction of workers are not gaining income or are even made worse off over time and when the economies think about what could be responsible for some of these changes then one of obvious stories is that we see a technological transformation that may work against workers a lot of media reports such as in these two american magazines shows uh pictures that allude to a battle between robots and humans and the robots coming here in form of terminators are winning this battle and are terminating eliminating human labor the argument that underlies these dystopian prediction is building on two different thoughts the first observation is that over time we have seen a very very rapid progression in computer technology that progression has led us to have ever more powerful computers it has led to a situation also where the cost of computing power has been rapidly declining over time a second uh evolution that we see is that we read almost every day about new uh new areas in which computers and robots are being applied the self-driving car of course the google car and center is perhaps the most widely debated case but a computer and robot uses are spreading in very many directions if he now combines these two trends cheaper computing power and more able more diversified uses of robots then we get to the following prediction we say that we're moving to a towards a future where we will have ever cheaper robots who are ever able to do a broader set of tasks so if you just extrapolate these trends far enough into the future you will eventually arrive at the situation where there is a cheap robot for every kind of work that one could do and of course employers will only employ human workers if the human workers are not more expensive than the robots so as the price of robots goes towards zero eventually the employers will pay ever smaller wages to the competing workers and when wages fall towards zero no one will be willing uh to take these jobs anymore and we indeed arrive at the end of labor so clearly as you can see this prediction is not totally uh out of this world because it just takes the extrapolation of strengths that we are already observing and we somehow arrive at that conclusion however this story is actually not the only story that you can generate when you extrapolate ongoing traits indeed you can also arrive at the exact opposite prediction using essentially the same methodology and this is how that works when we look at economic growth in the country that at a given point in time was the leading country in terms of technological development first the united kingdom and later the united states we see the following picture for a very long time economic growth was almost zero now what does that mean zero economic growth it means that our ways of organizing production were not improving much people did not come up or did not implement a lot of new technology that would have made us more efficient in producing goods and services so for instance if you went uh to a countryside here in europe in the year 1300 and then you somehow returned with a time machine in 1500 you would have noticed that the way agriculture is done is essentially the same and almost all of the economy was agriculture so not much changed even over hundreds of years of course now you could no longer say that if you just go back in time a hundred years agriculture even just visually looked very different from what it looks today so what happened what happened was that in the 18th century the industrial revolution started and this was the first time that we really had major technological breakthroughs that changed the way in which production was organized and allowed humans to use the same amount of labor and capital inputs to produce a greater amount of output that then led to an enormous spike in growth which reached its high point in the two decades following world war ii there are many countries including the united states as the technology leader but also many countries beyond that had phenomenal economic growth rates but when we look after 1970 we see that these growth rates are shrinking and suddenly it no longer seems so clear that we are able to produce that much technological change to really improve our production and make us more efficient of course you see that the growth rate is still quite a bit higher than it was in uh in medieval times but if you just visually extrapolate this line it could just as well be that by the year 2100 at the end of this century we will again be down to the type of near zero growth that we had for many centuries before so why could it be that the growth here is stagnating so much and how can it possibly be that there were earlier times where technological progress was greater well let me just give you one example which is these three innovations electric light the radio which is wireless transmission and the combustion engine all of these inventions came in one single year 1879 now imagine this in one year we invented a technology that we need for cars and many other modes of transportation in the same year we invented the technology that we now use for all of our cell phones and wifis etc and we invented the productive use of electrical energy which we use for pretty much everything today so compared to this our recent advances which are concentrated in the area of computers and robotics are perhaps not that big uh clearly having all these technologies here probably is more useful than having facebook and twitter but more broadly what we do see and what is a source of worry is that our production of innovation does seem to become less efficient itself one well-known example is our advances in medical technology this graph shows you with the green line the spending that pharmaceutical companies have for research it turns out that since the year 1970 we increased the number of researchers in pharmaceutical technology by a factor of 16. so we have 16 times as many people trying to find new medicines now you would think that with 16 times as many researchers working hard to find new medicines we would find 16 times as many new medicines but the blue line there shows you that today we need about eight researchers to get the same amount of output that one researcher was having in the 1970s so that means despite multiplying the number of investigators by 16 the number of new medical drugs that we develop has only doubled and this trend is also found in a range of other applications even for computer technology itself you might have heard of the famous moore's law which says that about every two years the computing speed of computers doubles and this has held up quite well over a long time however what is usually not mentioned when people point this out was that we arrived at this technological progress with a small group of inventors in the 1950s and 1960s and in the meantime we need much more researchers and much more investment spending and research spending to get that same pace of progress so it gets harder and harder to improve our technologies what this example uh suggests is that depending on the time series that we extrapolate we can get to very very different uh visions for the future there is not only the view that we could have these very cheap and uh and enormously able robots that do all of our work there is also with the same methodology of extrapolating ongoing trends a vision for the future where it just gets very hard to come up with new ideas because we sort of have already made much of the great inventions that humanity can think of and eventually technological progress will become much slower in any case the area that we can talk to at a more scientific level is not a very long-term prediction where necessarily we will probably far off with our precision but we do have some systematic evidence on what technological innovation automation does to workers when we look to the past where we have by now an enormous amount of experience not only about the technological innovations that took place but also about how these innovations were perceived by people at the time and how the eventual consequences of innovations turned out for the labor market i will discuss that with some examples from the textile industry which historically has for a long time been a beacon of technological innovation and progress when you produce textiles one of the very first steps that you have to do is to produce thread out of fibers and the way in which this was historically done like on the street ways and how it's actually still done in some poor regions of the world today is with a hand spindle and the idea is that the fibers are connected to this hand spindle that this woman has in her hand and then this spindle is put into rotation and while it rotates the the fibers are turned into a thread this technology existed for centuries until starting from the 11th century onward a disruptive new automation came to europe imported first from asia and that was the spinning wheel the spinning wheel was a very ingenious invention because here the worker would use the rotation movement of the wheel in order to make that turning of the fibers much much faster and much more precise and so the outcome of this machine was that this single worker could now produce as much threat as many workers were able to do before with the hand spindle and the reaction to this innovation was in many places very negative several major european trading cities including venice and paris and cologne simply paint the use of the spinning wheel and for those of you who are experts in ancient german language i also gave you a citation from the emperor of the holy german roman empire a strange name that i never quite understood when as a student of history so in the year 1685 the german emperor banned a machine that was like an early version of a loom for weaving he said in a document that is preserved today that the reason for banning the machine was that against the feeding of one person one would make 16 other people completely useless to society and these people essentially would no longer have to eat just for now one person having something to eat so the idea was really the same that we still have today if we can produce the same amount of output using much fewer workers then what happens to all these other workers who lose their jobs as it turned out this idea of banning the spinning wheel and other machines turned out to be a really bad idea because there were always some some clever cities for a while this was the city of basel for instance that actually did never ban these technologies so in basel everyone could produce with the spinning wheel they've produced a lot of threat with this mass production technology was really cheap and then they sold jeeps right to cologne and elsewhere and so all of a sudden these people in cologne all whose jobs one wanted to protect with the machine then actually lost their jobs simply because they were using an old non-competitive technology that could not survive over time another milestone of technology development which is often credited as the start of the industrial revolution is the invention of this machine the spinning chain in the united kingdom and as you can see it's basically a spinning wheel on steroids it's like a machine that has the same underlying technology but produces many threats at the same time and this then combined with other inventions there were mechanized looms for weaving and these new machines work was reorganized in big factories one invented technology to power these factories with the steam machine and suddenly what we had was an enormous mass production which again was labor-saving and used much fewer workers to produce the same amount of output and that again led to a lot of adversities in northern england and scotland especially there were big riots uh workers under the leadership of a mythical figure called nate blood main here were moving to the factories and then setting fire on the factories and then and threshing the machines so as to prevent factory industries and with the hope of preserving their own jobs the british crown then actually slashed down all these protests and killed a lot of these people um in a in a a time of unpleasant employer employee relationships but sadly these workers not only had a very bad outcome for them but it also turned out that this name of this leader nate lot became known since then for the so-called laudate fallacy like the luddite error and what people call the luddite fallacy is the observation that this prediction that automation will lead to a long standing unemployment that is continuously increasing as more jobs are being automated that prediction just did not turn out to be true and it's often times for us uh somewhat hard to understand why this kind of intuitive logic is not true that automation which clearly makes some workers redundant does not create much bigger unemployment and the reason why that is not the case is twofold the first part is a little bit easier to see when we have new technologies it very often means that new jobs are being created that are specifically linked to the new technology so in case of the big factory uh industrial revolution in england a totally new field was the production of machines all these spinning machines weaving machines steam machines they had to be produced and maintained by people so that created a new set of functions for human workers but there is a second element which is probably much more important and at the same time much more difficult to understand the second channel comes through people's changes in consumption patterns and the logic is this before the industrial revolution the production of clothes was reliant on a lot of manual labor it took very very very many work hours until we had produced a shirt for instance and as a consequence the price of clothes was very high most people at that time could only afford two sets of clothes one for the weekday and one for sundays but we factory production the price of clothes became much lower because we now had mass production and a mode of producing clothes which had much lower costs so for a family that meant that instead of spending a sizable amount of money to buy few clothes they could now either spend that same amount of money to buy much more clothes or else they could use some of that money to spend on other goods and services that they could not afford before because they had already run out of money after uh satisfying their basic need for clothes and indeed when we look at the evolution of uh spending over the last one or two centuries we see that there have been massive changes we now spend much more on many many things that for a long time were just luxuries for the reach for instance going on a vacation at the beach or in the mountains was for a long time just an enormous luxury for the very rich people like owning a private jet today is a very special luxury only for the very rich but eventually as prices of many things became more accessible people even those in the middle class could suddenly afford to spend on many leisure activities and that in turn has meant that we have created a lot of new jobs uh the whole leisure industry for instance is a very very good example of a big job creation over the last century of course or other areas that have been uh dramatically booming or also education or health services which are also areas where most people now have access to a relatively high level of services that people did not used to have a century ago now you may ask what i'm talking about here with these examples from textile does that still have some meaning when we are looking about these novel technological developments in the fields of computers and robots but it turns out computers and robots are actually not that new this magazine title from the german magazine der spiegel says the computer revolution progress creates unemployment and this magazine title is from the year 1978. it is 40 years old actually older than i am so when one actually reads that magazine which i obtained the experts in that magazine predict that western europe by the year 2000 will have an unemployment rate of 50 percent because 50 of our jobs will have been lost due to robots and computers and as you know while things are not looking all that great as i said initially definitely were very far away from an unemployment rate of 50 percent and the fact that this prediction was out there that people who argued to be experts said those things uh 40 years ago should give us some pause if we hear other experts making exactly the same argument today and saying robots will totally take over 50 of our jobs very soon now when i say that uh these predictions of uh robots eliminating work had not turned out to be true in the past that does not mean that robots and computers had no impact on the labor market instead work by my long-term collaborator david author of the massachusetts institute of technology and several collaborators including myself has pointed out that there has been an important transformation of our employment related to the different types of work that can more easily or more difficult uh be done by machines when we look at what computers do what robots do then we see that they have some very distinct advantages in comparison with us humans the first gigantic advantage of the computer is calculation that's actually why it's called computer for calculating things computers are also very good at handling information if we have to store information retrieve information send information process information all of that can be done quite quite efficiently and cheaply by machines these days the same is true when it comes to production processes in which we have a lot of repetitive motion that have to be done with great precision there again robots are in really their best uh application field but there are a distinct set of tasks that are quite quite hard to do for machines the first step is creativity reacting to novel things problem solving all of that is very difficult because there is no pre-existing solution and we don't even have sort of a good sense of what the space is of possible actions that could lead us to a solution so that is something that is very very difficult to comprehend for a machine and even ai experts debate on whether there is any hope at all that the machine could ever be doing this very complex decision-making test another issue which is difficult is social interaction with humans of course we know that machines are making progress there but it is still the case that it is an uphill battle for the machines as soon as it comes to more sophisticated interactions that need empathy that need us to convince someone then it becomes quite difficult to replicate that in a machine world a third area where machines are making substantial progress but still lag behind humans by far is when it comes to visual recognition figuring out what an object is and interestingly also fine motoric movement you heard on the first evening of that conference when uh worry here was i was talking to a robot that that robot while he was able to talk you know reasonably well still has some struggle like moving his legs or grabbing things with his hands interestingly grabbing things with the hand is super complicated for a machine a while ago i was visiting a plant of the of bmw the the automaker in germany and for the first two hours of the of the tour of the plane one sees almost only robots so that's all the heavy lifting right making the car body uh welding the pieces together painting the car all that is done by robot arms and then in the last hour everything is full of human workers and what they do in the last hour is they do the individualization of a car so the worker sees on a panel that says the person who is buying this car wants this navigation system and that radio and wants like green seat covers and a big steering wheel and then the worker gets all these things on a plate and then puts in all these pieces and the simple advantage of the human is the human worker can grab and put in many different objects or you could easily do a robot that can put in one object but a robot that can handle 16 different pieces that is just way too complicated and too expensive for the time being so what does this mean for different occupations on the one hand with computers and machines making their biggest advances in these information processing and repetitive production tasks it means that a lot of clerical work in offices and production work in factories can be automated at the same time there are skilled workers like managers or engineers or other professionals who are in their daily work very dependent on the flow of information and for them these new abilities via communication technologies offer them opportunities that make them much more productive for instance if someone is a manager of a big firm it has become much easier to keep an overview of a firm with worldwide operations because suddenly you know what is going on in your manufacturing plant in argentina you can see it on the computer screen in real time that's much easier compared to a time where you had to either send the paper mail to argentina to ask them what's going on or maybe making a very expensive phone call finally there is another group of jobs that is not that easy to automate i already told you in the car plans those people who put in the last pieces those are hard to automate but also many people who work in a restaurant many people who are cleaners those are actually jobs that take a lot of these visual recognition uh and find motoric movement and sometimes verbal communication uh contents that are quite easy to do for humans even if they have not studied in university and yet it's hard to do that for a machine so now let me give you some more specific numerical examples taken from the u.s labor market the chart here shows the number of people employed in a job between the early 1980s and a very recent time and we see here this job a typical conveyor belt production job with packaging goods here the number of jobs in the u.s dropped from 400 000 to 250 000 clearly as a consequence as well of technological innovation we have jobs like a bookkeeping clerk someone you know who makes relatively standardized payments that follow a well-defined scheme that is now something that in many companies is done by big software platforms and correspondingly the employment in that job has also fallen quite sharply over time but there is a flip side to this an obvious case of growing jobs is of course jobs directly related to these new technological developments software developer is actually now one of the biggest occupations in almost every western country this is just enormously important by now but we also see for instance more growth in jobs of certain types of managers and we see also a considerable job growth for instance among food preparation workers here you see that for them employment grew from about 450 000 jobs in the early 1980s to almost 1 million by now those of you who attended alan krueger's talk yesterday may remember that on his slide the food processing workers were actually listed as the most automatable occupation but it's not going on so some people claim that they see it coming but clearly this is not currently something that we observe now an interesting observation that people had is that these occupations which are expanding or contracting are concentrated in different parts of the income distribution what i did here is to rank all the occupations in the u.s senses according to their average wage so that we have the highest paid occupations here on the right side now it turns out that these occupations which arguably benefit from the new technology like the software developers or also the managers they are among the fairly highly paid jobs and the blue line here shows that their share in total employment is increasing over time the jobs that are being displaced the office workers the factory workers are actually not the lowest paid people in the economy especially factory workers for a long time been fairly well paid and has been an important source of income and middle-class lifestyles even for people with lower education these jobs which are concentrated in the middle or just below the middle of the distribution are now declining and then we have a second poll that is increasing at the very bottom of the labor market are the food preparation workers and the cleaners which are also expanding and therefore this pattern has been called polarization we see more and more workers in the best paid and in the lowest paid occupations because it so happens that the automated jobs are often located towards the middle now the same picture has also been replicated for many many european countries and this chart shows you with green bars that the fraction of employment of the one-third of highest paid occupations is increasing essentially everywhere the fraction of employment in the middle page sort of occupations with the red bars is declining in all places and the fraction of employment in the lowest paid jobs with blue bars is somewhat varied in italy actually somewhat declining according to that data but the general feature that we see in almost all of these countries is that the middle third falls more than the top third and the bottom now what does it mean for total employment this chart in a more academic style comes from a comparison where we looked at different cities in the united states that adopt computer and related technology at different speeds and we find that those places that use more computer technology actually see a small decline in the fraction of the population that is employed that it's the negative blue bar but the effect is very small and we cannot statistically distinguish it from a zero effect effects are a bit more negative if we focus just on older workers above the age of 40 here on the right side here the employment effects are more adverse and some people who look at data specifically on robot technology have also found negative employment effects in locations where a lot of robots are being used now how can the labor market adjusts to this change in the occupational structure i think the best form of adjustment is if young people move into the labor market in those areas where jobs are growing and therefore take on different jobs than their parents especially no longer replacing those jobs for which the demand is in decline to do so young people of course will need skills that are not that easy to automate need problem solving skills creativity communication skills and of course i.t skills more directly this process of the young people moving into different jobs than the older people is very much ongoing and one way by which we see that is when we simply look at how the average age changes of people in the different occupations again ordering them by their average wage levels what you see here is that over time over a period of 25 years the average age in all occupations is increasing because the population is getting older over time but the increase is especially big just left of the middle exactly in that area where employment is declining and the logic is simply when there are no longer many jobs in the in on the factory floor and no longer no longer many jobs in clerical occupation then young people are no longer moving there instead the young people are concentrating precisely at the top and the bottom of the distribution of course with the concerning consequence that this polarization of the labor market that you showed you before is even stronger for young people who are now separated into those that hold i.t jobs and the like those that work in the restaurant what is much more painful for the labor market to handle is if adjustment goes through unemployment so if we adjust to the use of machines by firing workers then it is oftentimes hard for workers to find new employment and that is even more strongly the case if you are looking at older workers who often face greater difficulties in navigating the job market an obvious help that we could and should give to people is some form of retraining that helps them find new jobs unfortunately we know from a lot of evidence that retraining is quite difficult to do we don't currently have a very good template for training programs that are very cheap and very effective this is still something that economists and policy makers are working on now of course one thing we have to ask when we see these uh these trends that replace the displacement of the so-called routine activities in production and clerical job and the growth of jobs at both ends of the spectrum we have to ask is this something that will continue in the future or will we see a new wave of technological change that is very different and there is some argument to be made that there is actually a bit of a paradigm shift coming about the old technology has as a critical feature that it operates with computer programs so the robot or the computer can only do what the programmer has told the computer to do that means it can only work if there is a well-defined sequence of work steps that a human initially told the machine to do and the new shift that is coming now is artificial intelligence and machine learning and the novel feature here is that this brings us the option that the machine no longer needs this exact description of the work process but we only give the machine the goal that it should reach and some general instruction on on how that goal could look like and then the machine computes the work steps itself using its enormous computation power one example of machine learning that is very prominent is visual recognition of objects and these two pictures are actually from a scientific publication that studies a pro problem that many people in information science study which is how do we recognize a chair and the difficult question here is well if you want to tell a computer what the chair is you could you could look at this chair and say you know the chair must have a flat part to sit it has a backrest has some arm armrests has some legs this one is a bit unusual because it has five legs so of course now if you take tell the computer that the chair has five legs then it will make mistakes it will not recognize this one on the left which only has four legs so we also have to tell the computer actually a chair could also have four legs and it could also have just three and maybe it has no backrest but then if it has no backrest suddenly the computer thinks that a sofa table is also a chair so then the program becomes more and more complicated and at some point you realize it's just almost impossible to tell the computer what the chair is so the solution with machine learning is instead we make thousands and millions of pictures of chairs and other objects and then tell to the computer look here are the pictures and we tell you which pictures are chairs and which ones are not now do compute which features of a photograph make it more likely that the thing on the photograph is a chair and we do have some fairly nice advances in that but nevertheless when we see these uh technological progress we see there are still errors you see one of the errors that this computer program made here it looked at this overturned table and thought it was a chair because he did not realize that this is probably not stable to sit on and also not very comfortable so when we look into the future i think that for now it still continues to be the case that any processes that are well-defined information flows in clerical work or repetitive production processes are the ones that are easiest to automate the new technology will expand and will ever more ever more tasks will become susceptible to potential automation but whenever i hear experts on artificial intelligence and the like speaking about what could all be automated i also think that we easily forget a number of obstacles to the implementation of technological uh innovation that come more from the side of economics and society and not so much from the technology itself the first issue that will become a big question for for us is how do we deal with machines that make errors now you saw the machine that identifies a chair that was actually not a chair that does not seem like a very grave error but of course we can ask what is happening now when the first severe accidents are produced by self-driving cars you notice we read about it in the newspaper so clearly people care about this issue to a different extent from what they care about accidents created by humans and this issue of the machine potentially producing errors is also one of the reasons why we don't have self-flying commercial airplanes despite having the technology of self-flying airplanes for over half a century a second limitation is that a lot of technologies have bottlenecks oftentimes it's not the computing power that really makes things difficult but there are other pieces that are really hard for instance for electrical cars the big problem is how to produce batteries which are light and small and cheap we have some advances there but this is really what's holding back the technology and you see that some technologies really improve over time but with other technologies we just never figure out how to resolve the bottleneck one example again from flight is the supersonic airplane we invented the concord airplane almost 50 years ago and it sounds like a great fling to fly to new york in half of time but we no longer use it because we did not figure out how to make this technology cheap enough so that it can be widely used and then finally one thing that is a great advantage of the human worker is that the human worker can do many tests at the same time and this is actually why i think that i would be much more positive about the employment prospects of the kitchen worker that we saw before which is growing a lot actually the issue is of course you can imagine that we could easily build some machine that cuts carrots for instance or and we have you know some of these machines to peel potatoes but of course the big advantage of the human kitchen worker is that the cook can tell you well do this do that do this do that run over there bring me that thing right so the kitchen worker just can do a lot of things once you think about all the robots and machines that you would need in order to do all these different things then suddenly becomes very expensive again to replace that worker and so i conclude on a rather optimistic note in answering my motivating questions clearly you feel that i'm i'm positive that our jobs will not be all eliminated by robots the challenge for us really is not one of uh of long-term mass unemployment the challenge for us is uh on the one hand how to manage this transformation where some jobs are becoming obsolete and people have to look for new jobs especially how we can help people who are becoming unemployed the challenge is to deal with the economic inequality that is generated by a trend where people concentrate in highly paid and lowly paid occupations and the challenge also perhaps might be that eventually we will not have sufficient economic growth to lift the life standards of a large fraction of our population and it is always important that we give people some perspective so that they feel that they can reach good outcomes in their lives thank you thank you david i think it was a very uh very nice and illuminating uh speech now we're going to open up the discussion we'll have questions from the floor please be brief when you ask questions the more questions we have the better there's a question down there i see a hand raised up can you please whether these trends that you've shown and the logic of different innovation could have a negative or positive impact on developing countries on the one hand yeah we can produce more more efficiently but on the other some workers will be replaced and does that leave space for the kind of infant industry argument and closing of technology things i think the the interesting feature of developing countries is that they are now going in that very different technological path than in our developed countries it's no longer the case that they are just like you know making the same transformation that we have only like 40 or 50 years later you see indeed that you know technology based on cell phones is actually often much more widely spread in those countries i would also say that a big opportunity for those countries is that communication technology makes it easier to provide services over the distance that that now allows people to uh to work towards the work process in the west operating out of a developing country what is certainly a big limitation in developing countries is always the lack of capital right one needs of course to set up infrastructure uh to uh to to be able to uh to operate to have uh computers and information technology but probably uh this uh limitation is actually less severe for these uh more service-based tests than any other questions can you hear me can you say okay right thank you now during the presentation you said that you were somewhat optimist but then you shared a slide that was not very optimistic strong polarization of your of work as you said and there will be people who will make it and will feel better and others who will not make it and will feel worse and will live in worse conditions now we have to agree about the percentages of those who will make it and this will not make it but then what can we do to try and you know fight this train if there is something that we can do to avoid this trend from occurring yeah so so of course the big challenge that we have is uh how can we uh provide you know attractive perspectives also for people who do not have a high level of education that i do think is a major challenge it turns out that you know for now in in some countries like in the united states there is actually a fairly robust demand for childcare workers restaurant workers and the like simply because people are able to spend more uh on uh on those services uh but i do agree with you this this will be a big challenge in dealing with uh what what countries are now trying to do for instance the united states is to subsidize those low-income workers with a program called earned income tax credit so the idea is that these people that if they produce some own income they get extra uh tax credits so basically a negative tax on top of that so as to reach a more political level of life if there is no other question maybe i can ask a question myself you said at a certain point that innovation may lead to new jobs or it may lead to other jobs associated with a change in consumption pattern but that is i think conditional on the fact that the gains that you have from technology are shared either through lower prices or higher wages uh in a sense somebody during this day said that the ownership of uh robots and um any machine learning so the ownership of this new capital is very concentrated and you do have some work also on the concentration of companies in the world so uh to what extent uh these issues are connected to the previous questions as well yes that's an excellent question i think there are different views on this uh there's certainly in in terms of distributional equity it's certainly desirable if the games are much more broadly uh distributed in terms of job consequences the patterns are somewhat less clear there's some people who say the very rich are exactly the people who then buy you know a lot of services and hire a lot of gardeners and and cleaners and drivers and so forth and therefore create a lot of employment but of course a really overarching issue is that the economy in many western countries has generated very big gains for for a small fraction of the population and the large fraction of the population has not done well and i think this is also a major reason why we may see a lot of discontent that expresses itself in elections which comes out of that observation that for many people the the economic progress has not uh been giving them their part of the games there's a question here can we have a microphone please so that we can ask the question here in the first row here at the center the gentleman with the blade jacket right here now in one of the first slides i saw a reference a possible reference set to the birth of a singularity how can this change the definition of the robot's characteristics not no longer robots that perform a standard tasks but creative tasks so singularity yeah so singularity is a is a you know very interesting concept based on the notion that at some point uh machines will reach kind of a intelligence computation level uh that is above humans you know all these predictions in my view that clearly go into an area where one has to say we just can't really know much about it almost by definition my view is that i'm just very skeptical whether you're really on that path because i do think that that a lot of this progress has become very very difficult to achieve and and i think my fear in a way is much less the fear of the robots becoming extremely good my fear is that we are getting slower than the current level of technological progress and if we get slower it will mean that for many people there are no income gains over time that we can currently envision because we just do not produce more overall income and then we're also in a bad corner of distributing honestly and that would be my main worry thank you we can uh raise the last question uh just wait a second for the and okay thank you see well thank you very much i have a question about speed now the human being now has a life expectancy of 80 years and works for 60 years the speed of technological progress uh is always related to the difficulty of human beings to be able to adjust to a workplace although you try it's very difficult to keep pace with technology unless you change completely yes it's true i mean this this is certainly uh you know very remarkable that we are now in a situation where over the course of a work life there is a considerable transformation um this clearly means that as workers we need you know to to be looking around and see what happens right try to acquire skills as time goes by but it also is the case that when we look back right on our parents for instance let's say my parents to get the generation here well those are actually that's a generation that uh that witnessed enormous amount of change uh it is sometimes a little bit i think the present bias that we think as young people my age that because we experienced all these transformations with computers and internets that we are unique and are the only people who had to deal with changes and i think the reality is there are certainly other cohorts that also had to adjust and ultimately what is important also in the labor market and what these values many places is people who just have a general skill of let's say problem solving right that an employer also feels you can hire someone who has not already done exactly that test but has shown an ability to just fit into a new job and figure things out quite rapidly so for instance in my country in switzerland most of young people do a professional apprenticeship they don't go to university and one very interesting observation is that later on actually a majority of these people does not work in the job they learned but they work in some other profession and part of this is simply that they have kind of learned some general skill on how to operate in a firm and how to interact with co-workers and likes and therefore their skills are still valued even if they're no longer doing exactly what they're initially trained for we have one very last question here well i have a couple of questions which are related one to the other now you appear that to be very skeptical when it comes to you know retraining programs for people who have lost their jobs can you give us some examples of countries that have had a positive impact in these so positive examples and then another question about skilled jobs that you believe are going to be more protected what can we do to you know retrain these people you know the profit professionalization of everything so how can we make you know waiters or cleaners and more professional yes so on the first one on retraining this is sort of a very very large field of study that labor economies are analyzing and typically they just analyze you know one program uh in one country in one circumstance and then the next one and then the next one and by now we have studies so-called meta studies right that that summarized the results of hundreds of such investigations and the result is just typically well sometimes things work but then you try to do something very similar somewhere else some other places only no longer works as well so the problem is just we do not have kind of this template right we don't have a clear knowledge where we can say we know exactly if you do this so it's great it's a bit different from you could say education where of course of course education is less than perfect but we at least know saying sending someone to primary school is a good idea right sort of learning how to read and write and do mathematics we know it's very valuable right that's those are skills that are super useful and and for retraining it's just more difficult now one thing that you point out which which i think is a very good one is certainly a hope for the less killed people is if you're somehow able to package more tasks into a job not quite sure how we can do that all that well for a cleaner but you can see for instance in uh in in a number of uh production jobs uh it makes for instance a big difference in production jobs whether a production worker is just pushing the button on a machine or is actually understanding the whole process process and some time ago i was visiting a factory in switzerland that produces plastic containers for yogurt i didn't even know that we still produce something like that in switzerland but it turned out that a the business model of the firm is that they deliver in real time so in at 8 00 pm in the evening all the supermarket chains tell them we need you know 250 containers of strawberry yogurt and then they produce it during the night and in the morning it's already in the store and the interesting thing there is the workers who work on these machines once uh the production phase is over they take out their screwdrivers and start working on the machines themselves and over time have been able to greatly improve the machines and the reason they simply they stand there the whole day and after a while they realize that there is a bottleneck that there is one part of the machine that slows down the process and then they have some idea on how to improve it and for that type of business simply the speed of getting things through the machine is really the thing which makes a lot of money and then of course as soon as you have workers who are sort of you know more than doing just a very simple thing but but understand things more broadly or in other areas can interact with customers as well can give advice then suddenly we have a package of skills that probably is harder to take away by automation well thank you very much i think we should give a warm hand to our guest i'd like to thank you all for being here very very active thank you very much david for this very nice presentation and i hope you enjoy the festival thank you very much foreign