Humans Machines and Work The Future is Now

Show video

You. So. It's my my pleasure to introduce Moshe. Havarti today from Rice University. I've. Known, Moshe, since, um, before. He was, established. As a researcher, actually his PhD, advisor. Was my first, major. Collaborator. As a researcher. At. The time he was at Hebrew, University and. He said you know I got this really great student, and you. Know you you, ought to keep an eye on and this guy is really going to be good I. Swear. That. His. Accomplishments. And. Awards read. Like a list of a, large. Fraction, of the awards that are open to any computer. Scientist. That. He's. Has three IBM, outstanding, Innovation Awards the cig act girdle, prize the ACM, connell aqus award the, ACM, sigmod. Innovations. Award Blaise Pascal, medal, that. I loaded a National, Academy of Engineering National. Academy of Science. American. Academy of Arts. And. Letters it's. So on the the. He. He, also has, done a huge service to the computer, science community by, taking. ACM. See ACM which was a really. Much. Ignored, journal. That. That. And. And, really, recast. It resurrected, it, really. Conceptualized. It as, in. A way at a vision for how it could actually be relevant, and widely read and he, actually made it happen and served in addition to everything else he does being a famous researcher in, several different fields and. Writing. Hundreds of papers somehow he managed, to carve out time in. For, ten years to to, lead that effort as, editor-in-chief so. I think we all grow Emma, Greta debt a debt of, gratitude. For. Having done that as well but today. He's on a new mission on. The. Affect that we have in. In. Automating. All kinds of work and just you know what what is that effect, first. If we're gonna if if we want to have enough. We. Want to improve the lot of people whose jobs are eliminated, I think the first thing to do is to understand, exactly what's, going on and he's. Going to be telling us about that today so no chef thank. You. Thank. You for the opportunity, today to address. You on this topic which is I think of the highest importance, and. Just to give you some idea where I'm coming at it so. It's. Probably fair to say for but still to you the filthiest for my 30. Years of my career. That. Was my main focus publish. Papers get. Them cited, get a good H number yada. Yada yada. But. About five years ago but. Five six years ago I start. Making and I'll explain a bit later how come into it thinking about the. The societal. Consequence. Of the technology that we all working so hard to develop and, I'm. Find myself about two years ago, in. A press conference and. A reporter from The Guardian, asked. Which, jobs will not be automated, and. I said would the general convey at the conventional wisdom is the jobs require a lot of a human. Interaction, and emotional, intelligence that these jobs would not be automated, and then. On the spur of the moment I said but I'm not so sure would.

You Bet against, sex robots. And. Of course this was a soundbite and you. See the Guardian run with this with this headline and. And, a. Thousand, newspaper quoted, this me on this one, so. Much so that some of my friends now call me sex robots motion. So. Let's, let's kind of recap the, AI story, first of all in a, very briefly so just to remind ourselves. You. Know you cannot exactly, trace exactly where it started you can find some people say it's not really in the 19th century you can go back and find people thinking about this but. Maybe, it's, it's not an arbitrary point to identify it with alan joy, who. In 1950, wrote a very. Well cited paper so, we are mostly, thinking about Alan Turing and Turing machines but. The paper that he published in 1950, Computing Machinery and intelligence publishing. A philosophy, journal mind I believe. Is his most cited paper and. Again. The paper is very famous for the imitation game the Turing test but, that's actually my opinion the weakest part of the paper the. The main part of the paper is a philosophical paper, on the possibility, of machine intelligence and, tooling. Goes through a list of objection. To machine intelligence and, he, decisively, unused, against, all of them and he, concludes by saying yes machines, will be thought, to be sinking he said by night by by the end of the century we, will. Talk. As if machines are thinking, well. The. Century has come and gone and I. Don't think still ever say well you know my laptop is thinking about what's the next slide to show I don't think that's how we talked about and. Indeed our. Optimism, was a common, feature in the day I in, the AI feel, you go back, címon. Renewal 1958. Well, chest 10 years max we will solve chess AI. Will, take us maybe, 30 years said. Minsky, 1967. So. Because of this early, optimism, there. Were periods we can own as AI winters where the, field have almost fallen into disrepute, when I was a PhD student AI. Was a suspicious, field you know people who over promise, and under deliver. Kind. Of a you can always come up with some names are associated with this error and. To. Me to me the turning point on a personal, level in. 1997. So I, was already a left IBM I was, already attached to university I have good still, very good connection with IBM IBM invited, me to. Come, and watch the game between Kasparov, and deep blue they. Paid my fly to pay for the hotel I couldn't say no. So. I go there and I, watch the first game in. The first game, Kasparov, was black, was, white which. Is perceived, to have an advantage and, Kasparov. Won, the first game, now. Remember then thinking okay I believe that one day computer will win but, probably the time has not come and we. Are still fairly new in new stone we didn't have many friends I left my wife behind I said okay I'm going to cut a visit short and I'm going to go back home and, I missed the second game, the. Second game the blue, was white. Kasparov. Had to play. More aggressive, game and he laid the trap too deep blue and, he, blew not only did not fall into the trap but. Came out of it with, a sequence, of move that was so brilliant, that. You see Kasparov, is reaction to that he. Walks off you know he he considered the game he walks off the game and for. Years he claimed that it was not really deep blue there, must have been a secretive, of gun master playing against him. He. Just came up he, just last year he published finally, a book to. The 20th anniversary of this when she finally yes it was the computer has won he, considered, it, but. That was kind of a really turning point because this. Poem is said that for years there were some time the future suddenly, it was the first significant. Accomplishment, by. AI and. Then. In 2011, I, came. Did it again this. Time with Watson and now. It was in jeopardy, and. You. Know we. All saw that the chess at the end of day that curve would change the computational, game if you can just look deeply enough into the game into the game tree you. Can you can defeat humans but. Jeopardy, require a whole different type of intelligence you, know you have to answer. Clever. Questions, about. Airport. In Chicago that, have some to do with world war 2 and you. Have to think well it's Midway Airport, I mean it requires, some all kind of knowledge of culture.

In History and language. Skills so that. Was the moment that for me I thought wow. It's. Not just about some remote future this is actually, happening. Now. I think the next, significant accomplishment, I would put what, happened last year which. Is that now, the mantle has moved to IBM. To Google, and alphago. Both listed all in go. Now. Go you, can see here go is in some sense a simple game you have a 19 byte by, 19 board just, whiten in black and white pieces there. Are many many more configurations. So. The game tree is known, to, be much, much much bigger and, the. For go he'll be considered a challenge after chess, people have tried many many years to do better in go and. Deterministic. Search clear was not going to walk so, alphago, used first of all randomized. Three, cells technique but still that's not enough so. Now the use machine learning and. Would have done first of all they digested all published games and then. Alphago stopped playing and guess it gets itself and built. A big deep model, of how. To make, what, a good configuration what are good moves. This. Is philosophically. Significance, because it means now that the. Way alphago, played is by having, deep. Intuition, not. Having, good rules how to play but having good intuition, and. To. Understand the significance, of this you we have to go back more than 15 years and. Find. A philosophy Palani Karl, Polanyi who ought in 1966. And this became known as a polanyi's paradox. How. Can we ever hope to have AI when, there are many many things that we can do but. We cannot explain how we do that if, you think about how. You drive. You. Just drive you, don't have a set of rules that you follow in, the, thought was if you don't have a set of rules that we follow it's. All about intuitive, feeling how, can we ever program, this in, answer is that we know what is dance right machine learning we're going to learn from data and. It's, interesting that the Palani put. Driving, as the kind of a a challenge, for AI because. It has been a grand challenge for right now for actually, more than a decade going, back in ninety in 2005, the, DARPA Grand Challenge. Eventually one by Stanford, well it's transfer team this is Sebastian tool in, more. Than 130, miles in unrehearsed rail in Death, Valley and. Two. Years later.

CMU. Now doing it in urban area and. Notice. This is more, than ten years ago this is before we had all the chips and all that we have today and before. We had deep, learning already. People were making significant, progress in, in. In. Automated, driving and. Of course by now this picture is I think, everyone here recognizes. There's. No clock called the Google car or more formally, the way Macau right. And, so. I. Think. This is an amazing thing was going to happen I'll talk more I want to talk about it because I think this is going to be the. Most significant. Technical. Development, that I'm going to see in my lifetime in suspect this is true for many of you okay, unless you're very young and then maybe you see other things but for most of us I think this is it and let me put it in a, context, why I think this is so significant, he's going to launch another transportation, revolution. What's. The first one invasion. Of the will but. 5,000, us BC what. Has he done to us well we'll. Not been able to build the pyramids if we didn't have wheels right you can't carry these big boulders, on your, back. What's. The next one after that the domestication, of the horse. Okay. Until there if you want to go somewhere you have to walk there now. You can rider you don't have to walk on your own feet and, if. You think about it you have the Mongol Empire which is the biggest land Empire, ever all. Conquered. On a whole spec. After. The domestication, of the horse which, happened, more. Than 5000 years ago. Nothing. Significant, happens in land. Transportation, I mean ignoring seed you, know water transportation. Until. The, steam locomotive, the same, locomotive, you can say it's the beginning of the enough industrial revolution in some sense in, a big way and we. Cannot imagine the United States is a continental, country if we, did not have this, trains, going. East, you know east-west before that if, you want to go from New York to San Francisco. What was the fastest way of getting from New York to San Francisco, a boat. All around America okay. So. The. Train the, train made modem on during the modern United States and, then. In 1908, the. Ford Model T the. Car at this point is about 50 years old always started in Europe in the, 1850s. I think but, this is the first mass-produce, mass, consumed, car and, it, gave us this. And. This. And. This. The. Car the automobile, I claim is the most significant, industrial, product of the twentieth century, we would like to think it's the computer, no it's not the computer the computer will probably be the most significant, product with 21st century but. If you bring someone from 50 years ago. 1967. And they. Go to sleep Rip Van Winkle, to wake up today how. Does the world look different well, every everybody, carries some something with them and a, lot of zombies are walking around looking at things but. Other than that the. Fabric, of life does not look that different between, 1967. And today. Some, years ago I went. To Cuba and, it was like going in in Time Machine 50 years back and. It looks dated, but, not dramatically, different. The. City in 50 years now in the future would be dramatically, different because. The cities of today were shaped by the automobile, they, were architect. By the automobile. They. Because. The United set is such a big country the, automobile you of such or such significance. It. Gave rise to a mighty industry, I mean. We can remember the chairman. Of GM. Saying 50 years ago what's. Good for Jim is good for the country. The. United States will not have become a superpower if you don't have the industrial mind it came from the automobile, industry. World, War two was worn by the Inc by the American industry. Think. Of how many movies the.

Ottoman Will play a role a cultural, symbol of adulthood. And freedom in. This is going to change. And. It. Will change for. One important. Reason and this. The cost the. Human cost of automobile. Is. Incalculable we just learn to accept it more than a million people every year are killed by automobile, crashes 30,000. In United States alone we. Just learn to accept it in fact even the old car accident, is a marketing, term, because. If you say accident. It's. A it's an act of God, you, know it's beyond us and, the. Source of the social cause of it if you actually go down to think about it how many people are killed, in Maine in effect, mostly people in the middle part of their life um about teenage. To-to-to. Early. Middle age in. Fact Texas only finally, now passed. The law it's. Now illegal to text and text and drive-ins in Texas, thank, you thank, you thank you thank you. The, leading cause of death among young people today. DW. Driving. While. Texting. Used, to be I right, now, it's about it's texting. So. This. I think this by itself justify. Pushing, for the, automation of car. Now. This is a book published by, by, Ralph Nader in the mid-60s blaming. The automobile industry but. The reason that we have so many car crashes, you. Want to know why look. To your left look to your right it's. All about us okay. 95% of. Of. Courses. Are caused by humans, I do. Sometimes an interesting experiment I ask everybody to write down whether. They're above average driver or below average rider. Then. I ask all the people who are above average to raise their hand and, it's, usually about 80% of the audience so. Probably. Today in. Silicon. Valley this, is kind of the hot one of the hottest. Technical. Trend, you. Have. You. Know some estimates, over 250. Companies working this area from. The from. The mainstream and car manufacturers, GM Ford, and the like. Some technology, company. Way. More Baidu. Maybe, up and nobody knows, the. Transportation. Companies, such as uber and lyft. Lots. And lots of startups. Technical. Issues there's. No agreement, some people say five years to me that sounds overly optimistic maybe. 15 is more realistic, the. Adoption, will be more difficult I don't think it will take 2025 years, for for, adoption will be gradual, than many many legal issues that need to be resolved in, many states the, law says a moving vehicle requires, to have a human driver in the car so, the laws will have to change it. Will be profound, business disruption. Most. Car seats idly. And doing nothing and I will have an opportunity to actually have a more rational and more effective. Use of the of the fleet. Huge. A major. Loss of business including. For example, people. Are now worried, where, will we get organ for transplantation. If. We don't have all these people king bike by car crashes. Still. I think of it is going to be a huge benefit to society mostly. If we just can save every, year million lives I cannot. See anything more, important, than that. But. Then let's look at this picture. This. Is the oldest. Say this is 2014. I don't think it sent automatically, it, shows all the 50 states and it shows the, most common, job in every state in. More, than half the states, it's. A truck driver. Now. This doesn't mean that the majority of the people in the center type driver this is an artifact, of how you aggregate, jobs. But. Still he tell you that lots of drivers turns. Out in the United States they, are about 4, million talking taxi drivers if, you, bought in the category, you look at people, whose. Job include, driving is a major component of the job think, of like postal. Workers. Now. I talked about 15 million people, and. In fact if you can automate, vehicles, you can automate really, the whole supply chain and. If you look for example at cargo, cargo ship well. They're much easier to automate, and then Karthik to deal with kids chasing, balls. On the on the street and, in fact the parties are going to launch this, year they are also planning to launch a cargo ship, if. You look at port today port. Today are almost. Hundred percent, containerized. And it means they can be fully automated I am, sure you have seen the videos. Of the Amazon. Warehouses, we have the robots that move move, the things what what still robots, have a hard time hard time doing is this. Picking. Up an object is still hard but they're. Working on it, so. We can see we, can see a new disruption, of the economy and massive. Lovel jobs. Now. You talk to economists about it and it's all yeah technology, destroy job it's create new jobs there's a new, quali was being reached nothing, to worry about that's.

The Standard line. That you hear from economists, here. Is a Ken, Rogoff a. Neoclassical. Economist, at Harvard I, think and, he says all people have been worried about mass unemployment since, on revolution it never happened, it never will. Now. Notice that even he says something about a, long. Period of painful, adjustment. Excuse. Me how long exactly. It, doesn't say let's. Move on don't worry, nothing. To worry about. Other. Economies, such as a poor Krugman, noble. Nobel, Prize winner and. Columnist. For New York Times is. A bit more, pessimistic. He said can innovation, and progress really. Help large number of workers, maybe. Even workers in general it. Was is that it can yes. Maybe how workers will be helped, so. There is this debate among, the. New Luddites, were saying, this time it is different and, they're. Neoclassical. You said this time it, is not different, so, who is right. So. This is now a matter in in in. Fact you can really given a popular media almost every week there isn't a new article about this topic and, in particular decide, new studies, McKinsey. Has a study governor, has a study Oxford University has a study, osed has a study past, water compras is a study everybody has a study. And every, study come up with, different. Projection, for the future. So. Why are we seeing so many, different predictions for the future, well. There is a Danish, proverb and he, says predictions, are hard especially, about the future, but. I think the Danish got it got it wrong predictions. Are easy especially. About the future everybody can make a prediction go. Ahead make a prediction. Quite. Predictions, oh this is the hard part okay, now, notice all this prediction usually, for. 30 years in the future, who. Is going to check in 30 years where the Gardner was correct or McKinsey, was quick nobody's going to check it and so. Our. Answer is nobody not what if what's in the future the future is inherently. Unknown, okay, we, can predict the weather about 24 hours into. The future that's about two maybe few days into the future, instead. I want to spend now the rest of the day a big part of the time talking, about the past. And. I'm going to use a case study. American. Manufacturing. Why. American manufacturing. One. First. Of all United States is a good test case for this because. It's a very large economy, it's about 25% of the world GDP and, it's a very market, based economy, more, than many, of the economies in Europe for example okay the government plays smaller role in the economy here then, in many many other countries a manufacturing. Is by. Far the biggest sector, of the economy almost, twice as big of the, next sector which is government and, government, here is government in at all level not just a federal government so. Manufacturing, is huge just, to, give you another, example. Of the size of manufacturing. US. Manufacturing, is, about, the size of Germany and Korea and France in Russia and Brazil and UK combined. Ok. China is slightly bigger the United States huge, manufacturing, is huge. This. Become a little bit as a surprise, because been hearing ah, manufacturing. Is all gone right, everything. Went to China in Mexico, but, this is all propaganda manufacturing. Is huge. Now. Here is to, me this is a slug when I saw it it, was a shocker. So. US, manufacturing. Look at volume. Inconstant, dollar in this disco, you can, see it's been going up for 50 years more than 60 or zero of course in zigzags there are economic, cycles. But it goes up but. Employment. Peaked around 1980, and, it's, been down roughly since end with, some a little recovery off from the recession here but. We have lost roughly. One, third of manufacturing, jobs over. The past 35. Years. Another. Way to look at it which. Is that a manufacturing, is a. Fraction. Of the economy, it's, very very stable it's. Around 12% and, it's, been like that for for, more than 50 years but. We used to have 25%. Of the of the of the. Labor force in manufacturing, and now we are at about 8%. So. Why is it happening well. It's very clear white is happening it. Is happening because, manufacturing, productivity, has increased, over. The past 20 years it is roughly doubled, in. Effect it's surprising, we only lost one third of the labor force in manufacturing, other than half of the labor force in manufacturing. So, the, point is if. Volume. Is up, now. What is the role of China here so, because of the globalized, world the, mark that in the manufacturing, worldwide. Is more, competitive, and what. It meant the u.s. had to shift, from. Low margin, industries, so, we don't make sneakers anymore in the United States this. Shirt was probably not made in the United States we. Move the manufacturing to. High margin, high, capital, capital intensive, industries, okay, so when I said that industry is going up it does not mean that all is that there is no huge change underneath an, automation.

Globalization, Are tied together and, automate. Automation what globalization, is done it has. Increased. The competitive, level and that, that enables automation. To become more profitable, automation'. Would have happened anyway, because even in this country it's very competitive, so even if we don't have to compete with China one. Company will automate it force the other company to to me. And. If you want to see automation, just, Google, excuse, me search search, Tesla. Model. S factory, floor and you can get very nice videos, of the. Tesla factory floor, where. They will tell you the only reason the lights are on so they, can take the picture, otherwise. Could be a dark factory floor and it. Reminds me of it used to be an old joke in GM when they started putting industrial, robots the, joke was the the, factory of the future will have all, robots. One. Man and one, dog and. The. Purpose of the doll of the dog is to make sure that the men does not touch the machines. And. All of the of the men is to feed the dog. So. Now. We can ask okay all this happen all these Mews of people lost their job, what. Happened, this is a good test guess what happened right so that the new. Liberal economist will tell you all well they should have fun other jobs there should be a neutrally balloon so, we now let's go and examine what really happened, and, in. A nutshell what, I want to show you is the last 40 years it's, been actually very harsh, on, working-class, and middle-class low, middle-class, Americans. And this, is something I have to say it's, the data that I share came as a surprise to me and I've, been reflecting on it why is that, I've. Concluded that we hear a lot about the. 1%. But. The real dividing society is not the 1% than, 99%, is. The 20% versus, the 80%. Were the 20%. All, of you are part of the 20% because, you are educated professionals, and. That's roughly people with undergraduate. And postgraduate education that's. About 20% of the population and. For these people if you, think about how you spend your life I would.

Bet The vast majority of your friends, belong to the top, 20% and. If. You have, a spouse most, likely your spouse come from the same group so, this group creates its own bubble. We. All live inside this bubble and everything. Is filtered through that bubble and we see very little what's happening outside that bubble we have little, idea what's. Happening in decaying towns in the south or in in the Midwest and. And. The bottom line is that I want to share another automation. Poses. A very very serious talent to developed economies we can discuss a developing. Economist if you want at the very end and I, think as a society we must develop policy, to address this challenge, so. Let's see what happened to to. The economy over the last 40 years, so. This is data that was put together by Eric bina from Andrew McAfee at MIT and, they. Have looked. At phone numbers that are very important, here one is productivity, because the economy goes either because, there, are more people or, people become more productive and, so. Productivity, growth is the engine of economic. Growth and indeed. You see here that from the early 50s for the next. 30. Years. Productivity. Goes and with, productivity, GDP. Grows the economy goes, jobs. Are generated and income. Goes, so. This is kind of the Goldilocks the Goldilocks, period and you look at all these numbers and you think ok, that's. So nicely, well aligned did you think there must be some economic law that. Productivity goes drive, everything, everything, goes up. Accepted. What happened, starting. From the roughly to her mid 1980, we see a divergence. We see a decoupling, and, he. Stood the day the, quad activity continued to growth and the GDP grows, but. Were not creating enough jobs and. Income. Has stagnated and. Here. Is an example of income, stagnation. This. Is a thing a nice way to look at income stagnation this, is real. Median, yearly income of white. Men with no college degree. And. You can see that it's actually has gone down between 75, and 2015 has gone down. And. This is I have to set a few slides over to me when I saw them I gasped because to realize that here in the United States, for. A very large sector of the economy people. Think have gone be now gone wrong now for 40 50 years kind. Of shocking and think. About what we saw in political, rally in 2016. We. Saw, angry. White men, okay. Now, we used to see this usually, there would be some of in the Middle East and they, usually we say something like death to America death to America, and now, we are kind of finding these people right here in our meets and you, ask okay what is so what are these four angry about, when. This. Is what they are angry about okay. That's it their life quality declining. For, whole generation. And. We. Also have been hearing a lot about about. Growing inequality. So. This shows you that if you look again this is between, a. Nineteen, eighty in 2013. And, you see the bottom ninety percent decline, this. Is the top one percent and. Then, you see the people who really did well out the top one percent of the top one percent they. Did really really well in this economy, and. In, particular, what it means is that. Part. Of the of the American Dream was of. Life, getting better except. What you see here is that the top one percent. Used. To own about twelve percent who used to get about twelve percent of the income it's, gone up to twenty percent the. Bottom fifty percent used to get about twenty twenty percent of the income, scan. Down, it's. Like some, people call in tale of two countries, where, one country is doing better and better and one, country is doing Wars and Wars. And. And, on top of it we used to think that this is a country that enables, of social mobility, that. Is to say that every generate the next generation will do better than the current generation and, what. You see here is that in, 1940. If, you're 30 years old the, poverty was ninety percent that you are doing better than. Your parents when there were ninety you went thirty years old now. It's 50/50 it's a contours, whether they're going to do better or wars the. New parents when there were a similar, similar age. You. Skip it okay this is another very. Significant, thing you hear a lot about how unemployment, now is very very low. Unemployment. That. Is reporting, the news it's. A charade. It's. Called u3, it's. A concept. Invented by politician, Ozzie to give us good news because, what does it measure how. Many people. Looking forward jobs are not finding a job it. Does not measure how many people stop looking for a job because they gave up and, this, was to me the second gasp. It's. The Labour Party's labor vote participate, these are people who are either working or looking for a job he. Said they would call labor force. This. Is for men between, 25 and 54 the.

Reason It's important to isolate, men for, two reason why the picture, for women you can imagine more complicated, the first entered, the workforce starting, from the 60s, now it's also in decline but, it's a more nuanced picture also. We're talking about manufacturing it's, mostly men here. You see that it's been going down now for. 50 years and. It's. Even starker when. You look what. Happened and you include a Duke education. Level you see the people whose bachelor degrees gunned down a little bit but. People with high school degree or less it's. Gone down even more significantly. That. Means that if you look now at, a. Men. With. High school education or less, probably. About something like one in four is not working. And. This is to me when I hello I said this is unbelievable, this is what we think of a country in, the midst of a depression when, you hear but Greece this unemployment level, of 25. Subpar, 25, percent not working, outside. Of the labor force this, is this is a I cannot. Think of anything listen this isn't to me a national crisis, except. If we don't learn about it I mean. Though if you dig in labor economy or say oh yeah we know we know about this but. This is not only knew that the 25 percent of working class men are not working working, class men are not working. Somehow. I've did a lot of hiding and nothing seemed to or cannot ever apologize, for that and the. Final thing that I want to show here on the label, is for. Many many years you saw that. Labor. And capital are fifth roughly 50/50 percent of the economy and. Except. Again and this was the case for many years and people thought again it's an economic law that economy have offered to be split between capital, and labor but we're, seeing again label now declining for, about for, about forty years labor was declining. So. What's going on I think, the, funnel. Default the following plot kind. Of explain what's going on this was put by David Otto with an AK on labor economist at MIT and what, he said let's look and here it's a combination of many many plots it's of a smooth, curve it, look at so-called the, skill spectrum, so. High-skilled jobs pay well and lost. Skill jobs do not pay well and this, looks if you look at the skill level our jobs, being generated, or declining. What. You see is in the upper part of the skill spectrum, we generate jobs ok. We, know that here in a. Computer. Science you know there the demand exceeds supply for.

Example This is a high skill job, at. The very end of the skill spectrum, we're, also generating jobs, why is that think. Of the. Job of cleaning table of the busboy cleaning table at the restaurant. So. First of all, you. Want a grand challenge in robotic, do. It build, a robot that can clean tables in a restaurant ok, we don't have the technology to do it the. Level of dexterity. Agility. And situational. Awareness that, the robot, will will will, will. Have to exhibit. It is way beyond what we can do but. We can tell take someone who is even not a high school degree and, pay. The minimum wage and they're going to do the job so there is no reason to automate, jobs and it added part of the spectrum so where do we automate, jobs in the. Middle of the skill spectrum. Why. Are we automating, jobs here take manufacturing, we, pay manufacturing. A employee. 20. To $30, an hour. Well. It's, a significant, investment ok, this is three times four times more than minimum wage, what's. The skill level it's medium it's a fairly routine job you keep doing the same thing over over, and over this, is only thing that we know how to automate, so, here this is the sweet spot for automation. What. David, Otto has shown auto, auto sure what happened over time and what you see is these, switch spots get. Border. Because. More a job is, the capability of forwards increased more and more jobs become can, be defined as routine. It's also move moves to the right moves, upwards, okay, so, this is like the ozone hole it gets bigger and bigger and. One. Way you can look at it is, by. Looking. At. Income. Of. Different. According to degree and you, can see the people have advanced degrees this. Is between 73, and now the income has risen this. Is true for so college if. You have just some college if you didn't finish four-year, you have some college yeah you you, you barely, kept up but. If you have high school or less you actually again you have lost ground these. Are the jobs that there. Are lots of people who can do this job so they're not paid well at all. Maybe. This is the problem here let me try tech meter right let me see what happened here. Play. Clay where is play. There. We go I didn't do that okay, well okay so, we have forgot to do that. Here. Is the the, final gasp gasp, slide this. Is data and covered. By. Economies to economists at Princeton they, look at mortality and, with. All the Borja run healthcare, mortality.

Gets, Keeps. Going down because health care it keeps improving, but. They found that this does not does not hold for white people in the United States it. Holds for Hispanics, hold for blacks you, can look at board it happens in every country health care gets better and better, but. They found that white in the United States our our mortality is going up this between 45. And 54, so this is middle-aged people and. Again. If you look at it by education. You find that this is a this is related, to lack. Of lack of a college education and if. You dig deeper and say okay what are these people dying from you, find the leading causes arm. Suicide. And. Drug. Overdose we all hear now about the opioid epidemic and, liver, diseases, and all. Of these were they call out this is of despair these. People. Are. Suffering. From despair, and. We can ask again what's going on I mean we're talking about right now for example I believe that opioid. Epidemic. Kill more people than, guns. Can you imagine what. Can kill more people gun in the United States. And. So. And. To. Top it this is women this is from March, of 2016. This. Is before the primary before there before even a trump, clinch. Denominator. Republican nomination, The Washington Post look, at the correlation. Between, mortality. And voting. For Trump and I. Found out that that, areas. Where there was higher mortality, more. Despair, too low voting for Trump, so, this economic. Development have, political, serious. Political, consequences. This, is what the economy the economists again put it put. Counties. By. Different. Index. Of Health metrics and, you can see if as you get healthier becoming. This blower and as, you get less healthy it becomes, red oh okay. So there is a clear, correlation here, between. Economics. And health and political. Inclinations and. To. Me this reminds of a very classic, quote I want. To see if anybody recognizes it. All. That is solid melts into error all.

That Is holy is profaned and, Men, is at last compelled to. Face with sober senses his, real condition, of life and his relationship, with his kind. Anybody. Recognizes, it. Yo'ii. I. Thought. It you'll have they like the Pope if anybody you should have heard the proper education. So. So. There is there, is a sense of dislocation that reminds, people with the feeling that happened in the middle of day of the nineteenth century, we'll, then I'll come back in, a few minutes to the Industrial Revolution, now. Look, at all this data and first. Of all urine has almost any economic, data people. Question, because, everything, for example you look at the I said, real, income, well, real income you have to factor inflation, how, do you measure inflation the different way to measure inflation. Every. Piece of data here is Conte is somewhat controversial, but I think the general picture will give something that that even if you need picot every single a plot, here you still see a general picture emerging. But. As everybody tells you. Correlation. Does not imply causation what, is the cause of this in economics it's very hard to determine causation. It's, very very difficult to do it so. This is a in. 2014. An. Opinion, poll was conducted the, Pew Charitable Foundation conduct, opinion poll amount about 2000, economists, and, again. It's, a poll so it does not tell, you truth we, have to take it with a grain of salt but. There. Was a. Significantly. It was a plurality for, accepting, the automation, is, a central, reason again it's not the only reason life is more complicated it's a central reason for, this for, the reason for wage stagnation. Now. I think the picture has changed and partly because there is a new study that just came up last year. From. Economists. At MIT and, they went city, by city and look at deployments, of an industrial robots, versus. Job losses and again. Remember. This all correlation. But, I think the weight of evidence is coming more and more that, automation, plays a major role in.

The. Dislocation. Of the working class people. Now. Let's talk about not this argument this time it is different or, not so, again the argument you hear a lot technology. Destroying, job but, it create new jobs, jobs that we cannot even fathom will be created. But. Let's, do a thought experiment if, you're working AI what is your goal your goal is to look, one. By, one at human capabilities, and try to test. A stream of artificial, intelligence try to be able to build, machine will be just as smart as people, let's. Suppose we succeed, I don't know this will take 20 years or 100 years let's suppose we succeed somehow. Because this assumption, means that there will always be something that people's are better it and. I. Don't see this why how you can argue that people will always be better than machine in fact unless, you, go into some metaphysics, we are some kind of a biological machine. You. Know evolution. Had millions. Of years to to, perfect, us it may take some time some time for us to to, compete with evolution, but I can't, see anybody who come up with an argument whereby all you people will always be better. Than machine, in some way I just don't see that. And. In fact if, economies, should know about anything it's, about the tipping point concepts as a tipping point okay, what is the tipping point it's, a phase transition there was popular book by Malcolm, Gladwell in 2005. Demonstrated. Many many cases, where a quantitative, change, finally, lead to a huge qualitative. Change and in. Fact what. Is the biggest. Qualitative. Change that we have experienced. In human history there. Are few you can certainly cultural, revolution then. The Industrial Revolution. So. Imagine you're an economist, in the Year 1700. And somebody. Asks you to predict economic, growth for, the rest of the millennium, we. Said well I have. Economic. Data going. Back since, Moses book that the two, tablets and this. Data shown me that, economic, growth is always maybe, point one percent per year so. We get confidence, I'm going to predict for, the next 300 years economic, growth will be point, one percent per year well. You would be wrong right, because. The Industrial Revolution happened, and so. Are. We now in the, in. On, the cusp of another kind of people talking about the force in a sort of illusion I don't know but, I don't know how we can through that possibility. And. In. Fact this, this, possibility. Was was a there. Was a very nice parable, attributed it appeared, in a paper by vasily lyon TF who was a Nobel prize-winning economist, and he, wrote about as a parable of two horses having a conversation, in. Maybe. In, 1910, and. One. Horse isn't is a new Luddite horse and he. Tells the other one other horse was neoclassical, have, you heard about the Ford Model T I'm. Incredibly, worried, I think I think this new technology, will destroy, jobs for horses what. Will we do after deployment, of the automobile, and the new.

Classical, Hall says come, on technology. Almost destroyed job for horses in it create jobs for horses there, will always be new job for horses jobs, we cannot yet fathom, and predict. Well. We, know which horse was right because, horse population picked, in 1915. And today. There's no job for horses, horses are pets this. Is this is the reality. If your horse we can argue that maybe it's a better deal for them I don't know that's a let's, say that's. A horse, equine, philosophy, debate. But. In terms of jobs there are no more jobs for horses. So. And then. People say come on yes not evolution happen and look everything, worked out we're okay in. Mind series really, have you learned your history of course we are okay now we're 200 years later but. They were, first. Of all to fully adapt when the Industrial, Revolution we. Had to invent, a new thing a new institution, it's called the social welfare. State it. Took 200 years, I mean the maturity. Of the social, welfare state happens. After, World War Two in the 50s and the 60s many, many things in this country only happened in the legislation, the Great Society administration. Of Lyndon Johnson, in the 60s so took us 200 years, to adapt, okay. It was not an easy adapt ation the Communist revolution, was about 100, million people who died in them even, in this country with for good there, were riots with. Dozens of people not black people white people being shot by police we. Forgot how the, how difficult, was to adapt to national revolution, and, now, we have learned nothing from it we're going to repeat we are going to wait for another 200, years of pain. And agony before we learn to adapt to the next, phase. Of Technology. So. Of, course technology, creates, open destroyed, jobs I mean. This is this is this is obvious, but. This is too simplistic will. Technology create enough jobs, how. Fast will it create them, what. About the skill level so take, a tag. Driver who. Lost his job because, now the tracks are being automated, and it's probably going to be the first thing would be automated would be long-distance. Tracking, would be the first thing probably to be automated before we do it inside, in the center of cities and. You tell this truck driver were, very sorry you lost your job but. We have openings, people, who write software for, autonomous. Trucks, please. Apply. Also. Where. Are the jobs being now, destroyed. In, small, cities in the south in the Midwest, what being created, lots, of jobs in in in, Seattle, ok, sell, a job sell, a house in Kalamazoo, Michigan and, move, to Seattle, easier. Said than done, people, cannot move location. Love location, location location was. Said about real estate and. Here. Is this, just came up on a but. A month ago on LinkedIn top emerging, jobs look at it list of jobs, they. Are exciting, list of jobs I, don't. Even know what is what do you. Unity. Virtual reality okay thank you okay how, many of these jobs goes to people with, high school degree.

So. We have an economy that creates. Fantastic. Jobs at the, high end of the skill adder. Well. You can do can you look at other cases but, I will tell you well then let's be come back to other type of jobs okay, so. This. Is very recent the oven it's Yanni down. The street from here that's an article in. The, in. Wired magazine Walker. DiSpirito, machine should try a new, job, caregiver. Well. Okay, how, much do we pay caregiver, again. At the high end if you're a physician assistant you'll, be paid well but that requires several years of training if, you're, going to go for, a low-level. Caregiver. Pays, very little because, the skill level is load a lots of people who come to the job does not pay well generally. People, form move. From manufacturing, to services, lose. 50 percent of their wages this is a big big, loss any of you consider, losing 50 percent of your income it's a big loss, but. Even, beyond the money, takes. This big guy who used to work on manufacturing, and we, said we needed to go and wash. Old people. And many. You can say well it's a noble job my, parents summer needs to care of my parents the answer is these people will tell you very often that's, not who I am and. Even. If they were willing to do that if they, if this, big guy would show up in a nursing home and the cells came here they. Probably he'll probably not get the job they probably hired a woman to do the job so, that cultural, biases, on all sides and, we. Can say that irrational we can complain about them, but, people are not widgets, people are people they come with all their biases, we cannot ignore it. And. We. Talk about quantity, of jobs, so. Here is to me this is very start go back to the talk just in 1990, and. You find an industry with. A market, value of about 65, billion dollar in real dollars and more. Than 1 million walk house now. Go to Silicon, Valley 2016. You, find an industry the big three, last company, which is which is a Google. And Apple in Facebook. And, the. Market value then was about half a one and a half trillion dollars, less. Than two hundred thousand employees. So. Something about the number of the new jobs mayor generate but the numbers don't match the old jobs at work that we are losing and. It's. Amazing that we went through election. Campaign in 2016. And. We heard a lot about trade, we held nothing about automation, not. From Clinton, not from Clinton and not from Trump I'm not surprised about time but I'm a bit surprised about Clinton the, only politician who need to talk about automation was. Obama, who was not running for anything and. He. Did observe, that we, are seeing jobs moving from. Manufacturing. To services, if, you are moving and then people are losing half, their income and. It says something which i think is very important, which is kind of get to the bottom line, everything. We know about work, in, our life is a fairly modern thing it's not just Industrial Revolution it's. Answer evolution plus, the social welfare state okay. And some, people call it the social compact that we are living by and it took 200 years to make it happen. Right. To strike remember. We saw the picture of the of people striking in in Pittsburgh. And we saw the train station burning and more, than 60 people shot by police the. Right to unionize abolish, child labor work, hour workweek, equal, pay for women minority, health and safety laws it, took a long time for this to happen and big, fights all over everything, because always there were resistance to every new improvement. In social. And labor legislation, there's. Now a new kid on the block called. Universal basic income and, universal. Basic income the idea is that we're, going to do away with a lot of welfare, program replaced by universal, everybody. Every citizen gets, a monthly check from the government. It's. A radical idea even. Though it goes back you can find Thomas Paine writing about it he said because, the land belong to everyone originally, there should be everybody should get some kind of a land dividend, you.

Find Interesting. You'll find support for this ideas from the left but also from the right people. Like Milton, Friedman, was in favor, charles Murray was in favor and what's there why do people you, can see why the Socialists, would support it why, would people on the right support it for, two main reason one, is they hate the nanny state they. Hate, the bureaucrats, making decision, about who should get money they said we're simple to get everybody, money and and, kill huge a part, of the bureaucracy, and. Secondly it. Will actually create a more efficient. Labor force, and the. Reason is the rebbe labor market will become more efficient because right under many people they, job sucks. They're not paid well they, cannot afford to, look for another job they have to be there they just don't have the flexibility, look for an. If, the heaven a safety network they, would say I hate this job I'm going to look for another job so. So, so, these are the arguments that people on the left and on the right they. Take this idea, one. Of the reason that I can, be skeptical about this idea personally is because it ignored the psychological, value of work it only deal with the socio-economic value, of or ignore, this the psychological. Value of work where. There is a lot of research at, least in today's culture it, plays such a huge role in our own personal, self, value. And. I want to finish with the court quoting a JSON for man who was the outgoing chair of Economic Council of Economic Affairs and Obama and. He says my worry is that, it's not that this time could be different come to AI, but. This time could. Be the same as what we have that. Is if you look at the past 50 years and, you. Just project, forward the same trend you don't have to change anything you don't have to say don't be another phase transition, just. Take it all this trend we have seen and just extend, them for the next 50 years and, we're in fairly bad shape so. I think we're facing to me that if you ask me that kind, of two huge policy. Challenges, and, domestic.

Policy Challenges. One is climate. Change and the. Other one is what I would call shared. Prosperity I, mean there's no question technology, will create a menace amount of wealth how. Do we do in such a way that the country is the whole benefit, from it thank you very much. Then. So. Go back to the horses, you. Know the, population, of the earth might decrease over time right, and so one possible, solution to this conundrum is, that the human population, just shrinks, do you think that's a reasonable. Possibility, and, if so could it happen fast enough that it would actually avoid some of the challenges. So. Let me board in your issue and I said the district ignore one huge, another huge trend in demographics. And. The more graphics is currently, one thing we see is ageing of the population and, so. Only, book says all we need all the orbits we can have to take care of the older people okay. This is one and the, other one is that as societies, get richer generally, the. Reproductive. Rate declines so. Demography, kind. Of think that we will peak. At about 10 billion then we'll start declining I, have. To say getting to ten billion they are talking about somehow the next like twenty one hundred that. To me is just too far to to really think, what happened I mean I'm just worried about how, do we navigate the next 25, 30 years so 2100. Is just a bit too far for me how. Aging, interact with it this is a huge question and I have to say I have not seen really really good answers about it so it's. A it's a very good questions what's going to be how, does how does technology and. Automation interact, with ageing of the workforce, I have. Not seen very very good answers to that I think nobody knows yeah, but, it's a good question yes. Speak. Up speak up so I can hear you yeah, so just sort of an observation so, you sort, of had on 2020 the idea of the universal, basic income but, the thing that you objected, to about it was was. That there's the psychological value. Of doing work so so, I wonder. You know how well studied, is it like suppose we doubled minimum wage right. Unemployment. Is very low right now undoubtedly, it would rise some, but. For the people whose. Have, remaining, work they would be getting substantially. More that might be a way of forcing the people that are making a lot more money right now to value, the services, that, those people are providing, more highly, and and. Maybe a way of spreading things out but maybe another way. Other. Than universal, basic income that still preserves that, aesthetic. That you. Like but so, so. A lot. Of speculation what, happened I mean look part of the issue is what happened when you, know right now our economy is based on people walking and consuming, what happen if more and more people aren't walking what will be the engine of economic growth and there. Are people who says well in a world where, you. Know most. Of the same produced by robots there will be suddenly a new human economy, just. Like that people are not willing now to pay more for handmade. Things, somehow there would be a flowers you know whole human economy that will create it i''m. Answer is nobody, knows you. Know this is the future you know nobody could have predicted a, hundred. Years ago how society, will evolve I think. We have a polymer hands now and, I don't know that we can come up with the ultimate answer, but. I think on the other hand ignoring, I mean telling telling, people don't worry you. Know if you tell people you. Know some people really look it's hard for you but your children will do okay maybe. People can take it if you tell people gonna. Be hard for you gonna, be hard for your children, maybe, your grandchildren will be okay for, most people it's not an acceptable answer right they would like to see so we need to come up to something, without worried, so much what, happened in a very long term answer is nobody knows would happen in a very long term, there. Was a question there in the back. So. I, have a question regarding the. Importance of the economy in the first place, some, some of the parties argue that civilization. Is not a natural state for human that. Actually together, is that's. What I want to be so. Couldn't, we have a different, different. Society or economy doesn't, play such an important role well, there isn't that psychological, needs to. Be acknowledged where work. Completely. Different system. Especially. When we have automation for support. So. Jared. Jared Diamond. Is. Famous for what his gun jumps and steel but, he also hold if you google you'll find an article he wrote humanity's. Greatest mistake. And. He's, argument you money greatest mistake was. Starting. The Agricultural, Revolution he, said we were better off as hunters gatherer and.

In A sense you can find the biblical, parable. Of the. Expulsion. From the Garden of Eden was. In some sense refer describe, what happened right so. They had to go and buy, this by the by, the sweat of thy brother, shall eat bread, so. You. Know you could be right but for. The last 10,000, years we've. Been living as farmers. And then as industry. And one. Thing we know, you. Know we can we may be able to handle slow, changes and we have adapted human human have adapted in an incredibly, resilient way but. We don't deal well with fast, change so. If you have suddenly you said we are going to go form between now and in 25 years to a culture of kind. Of enforced, leisure so. Some. People tell me our leisure is great I will finally be able to write all the poetry, that I want to hide and the, I always dreamed of taking a watercolour. Painting, and I want to then want. To do that we. Actually have data what do people do and are not working, an. Answer, it depends on their age all. The people watch television. Younger. People play, video games, so. Somehow we tell to me the future of humanity is TV and video games I find. It a dystopia, hey, I don't know about you yes. So, most, of the charter to shut off from the US yes, these trends, around technology are global, do you can. You cite countries, that be doing a better job on the policy front with. So. There. Is a whole, other talk somebody could give on the on the international, aspect, of it one. Thing that comes very clear when you look at international data is, policy. Matters, okay. How countries, respond, to that definitely. Makes it makes a difference for example Germany. Has. Germany. Is right now still has 25 percent of the people in manufacturing, but. Used to be 40 percent so it's going down but going down more slowly one. Word with one way they do that they're, working the working class people there and better education, than in United States so. This. Is something we're going to see how how. Different. Countries are responding, to that that's one one issue that we're going to we're, going to face yeah. There's. This gasp moment, which I first, like this we're. Developing this technology that is gonna generate this very painful potentially. Next couple of centuries. And. I'm trying to think what actionable, things we can do as as responsible. Scientists, to. I don't know the limit, what. Weird for building for building an atomic bomb. So. I'm really trying to, identify. Potential. Course of actions for us to, help, out we're, not economists, on this or not so. From the point, of view of machine learning doesn't. Make sense we are more. Thoughtful. So. You. Know that, they follow, the question is what should we do and who has the knowledge to do it and the answer is right. Now I don't think the knowledge is coming for us okay. We're pretty. Much I look around me we are techies and, they. What. We need to in fact we tend, to typically. We end up being a bit some of dismissive. Of social science but some, of this answer will come from social science they. Need to work with us partly, because they need to have an understanding where technology, is going what are the trends but. The answer is weather which, would have universal base income or not. This, is something that the. People, from computer science will come up with the answer. Max. Tegmark for MIT who is now very concerned about social, impact of Technology said the answer is not. Less technology. More. Social wisdom in the which the way he described, it. You. Know I've been to you know everybody's looking for some we. Are again we're techy we're like one magic. Bullet will, solve the problem answer, is we are not finding it we find a plethora, of things you know we can look at for example think tuck poor tax policy, so, we just said the tax reform. Tax. Reform give incentive, for, capital investment, is. That what we really need right now maybe we need investment to for, companies, to invest more in the workforce so. So, right now we just passed a new lesson about taxes and this encourages.

Capital Investment, and because. You can expense it immediately and less. Let's, say investment, in the workforce so even tax tax legislation. Is, relevant, to this debate what, about education, so right, now we are that, the education. System was designed if you finish high school you. Are ready for the job you're ready for a job it. Doesn't work anymore okay, we have to have to rethink our educational system so the main memory may be may be moving pieces, and. I'm I have to say I'm struggling myself what's, what's. Our position, on this what do we do in this thing and first, of all I would say is first. Of all we need to assume responsibility right. Let's we are this. Technology, it's us we've created it, it's. Not clear that immediately, we will say that we need to do something but first of all we should say we. Are part of the conversation we cannot ignore it yes. I'd. Like to see a chart that looked, at across. Our a lot of countries the. Proportion. Of the population out. Of work and the likelihood, that that country would have some kind of significant, revolution that, would destroy. The society. So. That's a danger, we. Might not be able to fix the problem before something bad happens, so a, handle. Mcafee for MIT his, quote is the. People will rise, before. The machine will rise and, so. Historically when, you look at countries where the, gap. The the inequality, grows more and more and more yeah. These are the thing that leads to very. Significant. Political. Outcomes. I think you look at what's happening United States now and you says well I don't, know if you call it a revolution but it's a very significant, outcome of what happened in in. 2016. Then. What. Technology. Is. A. Place, for the government to slow down say the adoption of automated. Self-driving, cars, for, example because, you know it would give society. More time to adjust right it's some level even, if we can do it is it the right thing to allow the. States or cities I. Mean this opens, a very, big question but technology and regulation, so, in this country who have basically said. The. Marketplace, will decide which, technology, are adopted, which kanojo developed we, are the government just steps aside there. Are some cases now I think that it's it's we. Need to really seriously question, that I would say the area where I see it is clearly. In terms of cybersecurity. You. Know we when, the automobile was introduced, people, start bringing all kinds of measures there was the National Technology, Safety Board it's from the nineteen twenties people, says well you. Know technology people are dying let's do something about it we. Cyber with cybersecurity. It's. Just people, are selling well it's like force of nature nothing we can do. To. Meet this we are waiting I think what's going to happen it's, going to be some, disaster, some real disaster, and then people's all we need to do something until then I don't, see anything happening you. Can look at other thing like crypto currencies, does, it make sense what are the risk of cryptocurrency is again we're just we. Just unleashed technology, and we let it happen and let's. Go see what happens I. Think. We're going to see I mean look well the situation now that you have a. Small. Number of companies with, a I, think the the big tech companies that will have market, equity, of close. To three trillion dollars, they'll. Call you a lot of the small number of people. Not. Nearly name some of them are paying very little taxes, you. Kind of put everything and you can imagine that Washington, would start looking. Somewhat. Askance, had taken maybe, I mean I said the same thing I have to say regulating, regulating, takers we've so it's not in their DNA. But. I cannot you know it's quite conceivable to me that somebody will start, putting various kind of legislations, about technology, I don't, think they would slow down the the necessarily, the flow of energy that's, the development. Of Technology but, there was amazing piece of news just this week Apple. Investors. Want. The company to do something about the. Addiction, problem among. Among children and teenager, for in iPhones. But, this did not come from the association, of distressed, parents is, that Apple, investors. Wow. This, is an aha, moment right, he some investors, are starting to worry this, is bad for business a, Mark. Zuckerberg just announced is going to change Facebook it's going to be a false for the good whatever, I, think.

2018-02-11

Show video