The robots or machine intelligence. The longterm consequences of the inevitable rise of the machines will be the topic of tonights lecture so pleased i may in a very autistic enthusiastic welcome for our friend martin ford. A. [applause] thank you very much and thanks for coming. Its really a pleasure to be here. I should say right off the bat i noticed in the program the title for tonights presentation is how to stop robots from stealing jobs and i have to tell you honestly i dont have an answer to that very i believe it is probably inevitable that technology is going to replace former workers and take over the economy and a challenge for us is really to figure out a way to adapt to that and to make sure that we still have continued broadbased prosperity in the future even as that unfolds and thats really the main point of my book. What im trying to do is to interview two or maybe even you might say initiate the conversation that i think is going to be critically important for all of us probably over the next couple of decades. I would say its a pretty good bet that we may well be on the leading edge of a massive disruptive way thats going to unfold may be over the next 20 years or so and its going to put a terrific amount of stress on society and on the economy. And the central idea here course is that machines are robots and smart algorithms are increasingly going to take over more of the work in the economy. They are doing things that people do and perhaps most importantly they are going to start taking over the things that people are paid to do great i think that could ultimately create a big rovlin for us. There have been a number of attempts to quantify all of this mostly undertaken by people in the academic world and they have set out to try to come up with some kind of a reasonable estimate to just how many jobs might ultimately be threatened in the numbers that they have come back with our pretty frightening, pretty harming really. Depending on the country they look at are the assumptions they have made the estimates have come back in the range of 30 to 60 of all the jobs in the economy. The most highprofile here in the United States was undertaken by researchers at oxford university. The number they came back with was half of the job that the United States could be susceptible to automation over roughly the next two decades of thats a pretty scary number. And the approach that they take or the assumption that they make when they make these estimates is to start by saying if the job is on some level fundamentally routine and repetitive and predictable than ultimately it seems likely the robot or an algorithm or some kind of technology is going to be able to automate that job. The word that is very often used to describe the jobs that will be susceptible to automation is routine and i think that can be misleading because it implies very often a job that is repetitive. If you look at the technology we have already its pretty clear we have gone far beyond simply automating Assembly Line type work. We are already well beyond that. I think a better descriptor word is really predictable. If a job the types of things you do can be predicted based on what you have done in the past than that job is likely to be susceptible and that sort of captures what is the Central Technology that is driving all of this. Machine learning is essentially having an algorithm shone through data and that might be Historical Data or Realtime Data from something thats happening right now in the real world and basically the algorithm figures out how to do things. The idea is you given al gore the man outcome that you would like to to achieve and it figures out for itself how to get there. In essence you could say its a way for computer to program itself as opposed to having a person sit down and program at step i step. Thats obviously an important change from what we have seen historically. One way to think about it in terms of whether a particular job might be susceptible is to ask yourself cut another smart person if they had a very detailed records everything you have done in the past eventually figure out how to do your job for the could they figure it out by watching you work . If the answer to that is yes i think theres a good chance a smart algorithm may be able to do the same thing. If on the other hand you are michelangelo and creating something generally knew all the time at least for the foreseeable future, i would say never i would never say never but i would say your job is relatively safe. Those kinds of jobs and the number of people paid to do those gentlemanly creative jobs is a very small fraction of our workforce. That doesnt necessarily offer us a safety net though. So if you really sit down and think about what it would mean for anywhere from 30 to 60 of the jobs in the economy to evaporate over period of perhaps 20 years its really hard to imagine too many things that would have a bigger impact on society and on the economy than that and the things you cant imagine are uniformly bad. There are things like wars and plagues and this is a big issue and something well worth giving a lot of thought to and well worth having a meaningful public conversation about. I think we really need to start to wrestle with this and figure out if indeed this is something thats really going to happen and what we are going to do about it. That has become a really critical question for us. So i thought before i get into talking about the book itself let me tell you what little bit about my background and how i came to start digging about this in writing about this. Back in the early to mid1990s i started a very Small Software company in Silicon Valley and by small i mean it was in my apartment. The First Software product was a tool for windows or grammars and Microsoft Windows was just taking off. As iran this company one thing i discovered was running a Small Software business back then was a laborintensive process. Software was shipped on physical media on cdrom. If you are going to sell a commercial Software Product and people expected it would be accompanied by a printed instruction manual. There was a lot of mandible routine work involved and when the customer ordered a software you would pack all the this stuff in a box and ship it out. Eventually as the business got bigger i ended up outsourcing back to another Small Company that specialized in Billing Software for Small Businesses like mine. They hired all kinds of people not necessarily high skill levels that did that kind of work to ship it off and handle calls from customers and that type of thing. So there were jobs for average people but as it turned out those jobs arent there for long. Within a few years the business basically change dramatically. Obviously now software is primarily delivered over the internet. In many cases Software Products are just posted in the cloud and customers access them remotely. The business that i helped source that work to actually would not dismiss by the early 2000. Much of that work had evaporated so i had an opportunity to see that unfolding in my own business and of course i was close to the technology and computing speed and how Software Development with changing and how was getting easier and easier to develop more and more sophisticated products. As i saw all of that it became obvious to me that what i was seeing in the very often stories about robots and technology and how they are impacting jobs. This is an issue that has gotten a lot of attention and i think thats a good thing. Its something that we need to wrestle with. This is a book that i think covers the fair amount of material. If you read it you hopefully wont come away with the impression that its one of those books that could have been a magazine article. There is a fair amount of stuff in there so what i want to do is touch on a few of the main ideas and some of the highlights of the things that i think are the most important. One of the first things and maybe the most important thing i do the book as i try to take on the question that is central to this and its a question that comes up more often and that question is why would this time be different . This whole idea that technology can displace workers and potentially create unemployment is a concern that goes back at a minimum 200 years. It goes all the way back to the revolt in england in the early 1800s. Since then its been raised again and again. It has come up any times and each time it has turned out to be a also alarm. Theres a long record of this alarm being raised and ultimately it always turns out to be false. I think there is a lot in common between this and the story of the little boy who cries wolf. In that story people eventually become very skeptical because the false alarm keeps getting raised and we become very complacent. In the end of course the wolf does show up and it doesnt go well so i think this issue could end up more or less the same way. Thats the primary concern that i have here. Clearly this is an issue that kind of sits at the intersection of technology and economics. Economists are very skeptical of that. Whereas with technology this time its always different. Thats the whole point. You are always going to place that no one has gone before. One of the most basic questions you can ask is if this about economic sources about technology . Ultimately its going to be about technology. I dont think theres any other mental law of economics that says people have to be people are essential to the production process. I think there is is quite a bit of evidence for that already. If you look at the new types of products and services that people are demanding that they want to spend their time with, things like spending hours on a smoker to it or playing video games and in near future they will be spending how who knows how long and virtual reality. Those are all services that are delivered essentially without people being in the loop. There is a role for people in their initial creation and the creation of content. Once the content is being created than delivering that service or simply isnt any labor content to it. Its all computing facilities essentially doing the work and i think that offers a preview of the way that things are going. I dont think theres any rule that says people always have to be essential to the production process. Its entirely possible that at some Point Technology can reach a point where there isnt much of a role for human labor anymore. Its understandable and people are skeptical because that is never happened but there are certainly instances in history that you can point to as well. Theres a new book out that selling very well and if you read that you will learn at the time of that first people were enormously skeptical. They were very smart and prominent people who said anything and airplane was never going to be feasible. Some people said it would never happen in a thousand years and others said even if it was accomplished it would never be practical. Obviously that turned out to be dramatically wrong. And yet its really difficult to be hard on the people who made that assumption. There was an enormous amount of data and evidence to suggest that simply was never going to happen. Basically people dont get into heavier than air contraptions that fly through the air. There were plenty of failures, spec secular failures to support the idea that simply wasnt going to happen. I think that sort of offers a preview of the way things are. Eventually we will get to the point where technology is capable of doing the vast majority of things that the average person is doing and that will have a very dramatic impact. In terms of articulating whats different this time one thing you can do is you can look at the example that is cited most often by the skeptics and that is agriculture. It used to be in the United States the vast majority people worked on farms. Now almost no one works on a farm. I think its less than 2 of the work wars works on a farm. Clearly that hasnt turned out to be a bad ring at all. Food is cheaper than is to be a people can move onto other roles, more fulfilling. It turned out to be a good thing so the skeptic will ask isnt that just going to happen again to . Yet if you look at what happened in agriculture clearly there was a very specific technology that impacted the Agricultural Sector and not the entire economy. What happened was millions of people did in fact lose their jobs on farms but then they moved to other sectors of the economy. They moved her stew agriculture and then to the service sector. The interesting thing to note is the fundamental nature of the work they were doing didnt change much. They were doing relatively routine repetitive work on farms and later they were doing routine repetitive work in factories and nowadays people do relatively routine work in the service sector. You might have someone working on a farm in 1900 factory in 1950 and at walmart scanning bar codes or Something Like that. Fundamentally thats the routine and repetitive work and all of that will be susceptible to these technologies Going Forward forward. You can point to a think really three things that defined Information Technology today and makes a fundamentally different than the things that came in the past and the first is acceleration. In general on im a fairly broadbased basis Information Technology is accelerating rapidly and its been doing that for decades. What that means is as you continue to double something again and again and you keep doing that over period of decades to get to the point where you are moving in absolute terms at an extraordinary rate and that is where we are now. That is why things are often developing at a surprising rate. We have been going to this redoubling process that the amount of progress we make is the second thing is that this technology for the first time on some level encapsulates machine intelligence. Its not like the tractors and plows and the harvesting equipment. We use technologies that on some level can think and solve problems and make decisions and most importantly they can learn the data and learn from that. Its very different from mechanical technologies that transformed agriculture. Third and most important is broadbased. Its not specific to any one sector of the economy the way that Agricultural Technology was. Its everywhere and its going to invade every business and every organization every industry and that includes everything that exists today and perhaps more importantly it includes all of the industries and businesses and employment sectors in the future. What that means is people will talk a lot about Creative Destruction and the fact that old things are destroyed in new things are created and thats absolutely true but i think theres a lot of evidence to suggest while there certainly will be new industries created in the future nanotechnology synthetic violet shade virtualreality which will all be very important future but its hard to imagine any of them are going to be laborintensive. I dont think any of those industries will employ huge numbers of people. We see that happening already so the real risk that we face Going Forward is processing Creative Destruction and the destruction will fall on her laborintensive industries which right now is in areas like retail, fast food hospitality all those areas that employ huge numbers of workers. Jobs will be destroyed in new things will appear in the future but they simply wont be laborintensive. Over time they will find it harder and harder to employ the work horse. Those are the reasons why i think this time could be different although certainly people remain skeptical. The second thing i say thats important that i focus on a lot in the book is that our conventional view of which jobs are likely to be automated is not quite correct. Attrition traditional view has been that the robots come after relatively unskilled jobs and the solution to that is to send people back to school and give them more training so they can move up the skills ladder and do a job that requires more of an intellectual input. Very often you might have a person who loses their job and a factory or warehouse and they send it back to school and perhaps they can find a job in an office and thats the way things are supposed to work. The problem is what we are seeing quite clearly is many the more skilled jobs are actually at least its easy to automate and did many cases more susceptible to automation than the more skilled jobs and thats especially true when they are what you might think of as midrange knowledge type jobs the kind of job where you sit in a cubicle at it computer doing some relatively routine formulaic analysis producing reports and that type of thing. Those jobs are going to be highly susceptible to this and the reason is they are fairly easy to automate. It only takes software to automate those jobs. You dont need mechanical contraptions are robotic arms or any of that expensive stuff. All it takes is software and Programming Software using Technology LikeMachine Learning to figure out how to do those jobs is in many cases quite straightforward. On the other hand many of the lower skilled jobs rely on things like visual perception and dexterity and building a robot that comes close to replicating what a human being can do in terms of dexterity and the ability to perceive of environment visually and manipulate that environment is beyond science fiction. We are Getting Better and better at it but there are still many jobs that are far below on some of those are good jobs. An example would be nursing. Nursing is a good bet. I think it requires a tremendous amount of dexterity and mobility and also it requires a high skill level in problem solving. On the other hand about the jobs are not that good in a good example would be someone who is a Home Health Aide for example to assist an elderly person or to as