Well, Arthur C. Clarke, a famous science fiction writer from the 1950s, said that, "We overestimate technology in the short term, and we underestimate it in the long term." And I think that's some of the fear that we see about jobs disappearing from artificial intelligence and robots. That we're overestimating the technology in the short term. But I am worried whether we're going to get the technology we need in the long term. Because the demographics are really going to leave us with lots of jobs that need doing and that we, our society, is going to have to be built on the shoulders of steel of robots in the future. So I'm scared we won't have enough robots. But fear of losing jobs to technology has been around for a long time. Back in 1957, there was a Spencer Tracy, Katharine Hepburn movie. So you know how it ended up, Spencer Tracy brought a computer, a mainframe computer of 1957, in to help the librarians. The librarians in the company would do things like answer for the executives, "What are the names of Santa's reindeer?" And they would look that up. And this mainframe computer was going to help them with that job. Well of course a mainframe computer in 1957 wasn't much use for that job. The librarians were afraid their jobs were going to disappear. But that's not what happened in fact. The number of jobs for librarians increased for a long time after 1957. It wasn't until the Internet came into play, the web came into play and search engines came into play that the need for librarians went down. And I think everyone from 1957 totally underestimated the level of technology we would all carry around in our hands and in our pockets today. And we can just ask: "What are the names of Santa's reindeer?" and be told instantly -- or anything else we want to ask. By the way, the wages for librarians went up faster than the wages for other jobs in the U.S. over that same time period, because librarians became partners of computers. Computers became tools, and they got more tools that they could use and become more effective during that time. Same thing happened in offices. Back in the old days, people used spreadsheets. Spreadsheets were spread sheets of paper, and they calculated by hand. But here was an interesting thing that came along. With the revolution around 1980 of P.C.'s, the spreadsheet programs were tuned for office workers, not to replace office workers, but it respected office workers as being capable of being programmers. So office workers became programmers of spreadsheets. It increased their capabilities. They no longer had to do the mundane computations, but they could do something much more. Now today, we're starting to see robots in our lives. On the left there is the PackBot from iRobot. When soldiers came across roadside bombs in Iraq and Afghanistan, instead of putting on a bomb suit and going out and poking with a stick, as they used to do up until about 2002, they now send the robot out. So the robot takes over the dangerous jobs. On the right are some TUGs from a company called Aethon in Pittsburgh. These are in hundreds of hospitals across the U.S. And they take the dirty sheets down to the laundry. They take the dirty dishes back to the kitchen. They bring the medicines up from the pharmacy. And it frees up the nurses and the nurse's aides from doing that mundane work of just mechanically pushing stuff around to spend more time with patients. In fact, robots have become sort of ubiquitous in our lives in many ways. But I think when it comes to factory robots, people are sort of afraid, because factory robots are dangerous to be around. In order to program them, you have to understand six-dimensional vectors and quaternions. And ordinary people can't interact with them. And I think it's the sort of technology that's gone wrong. It's displaced the worker from the technology. And I think we really have to look at technologies that ordinary workers can interact with. And so I want to tell you today about Baxter, which we've been talking about. And Baxter, I see, as a way -- a first wave of robot that ordinary people can interact with in an industrial setting. So Baxter is up here. This is Chris Harbert from Rethink Robotics. We've got a conveyor there. And if the lighting isn't too extreme -- Ah, ah! There it is. It's picked up the object off the conveyor. It's going to come bring it over here and put it down. And then it'll go back, reach for another object. The interesting thing is Baxter has some basic common sense. By the way, what's going on with the eyes? The eyes are on the screen there. The eyes look ahead where the robot's going to move. So a person that's interacting with the robot understands where it's going to reach and isn't surprised by its motions. Here Chris took the object out of its hand, and Baxter didn't go and try to put it down; it went back and realized it had to get another one. It's got a little bit of basic common sense, goes and picks the objects. And Baxter's safe to interact with. You wouldn't want to do this with a current industrial robot. But with Baxter it doesn't hurt. It feels the force, understands that Chris is there and doesn't push through him and hurt him. But I think the most interesting thing about Baxter is the user interface. And so Chris is going to come and grab the other arm now. And when he grabs an arm, it goes into zero-force gravity-compensated mode and graphics come up on the screen. You can see some icons on the left of the screen there for what was about its right arm. He's going to put something in its hand, he's going to bring it over here, press a button and let go of that thing in the hand. And the robot figures out, ah, he must mean I want to put stuff down. It puts a little icon there. He comes over here, and he gets the fingers to grasp together, and the robot infers, ah, you want an object for me to pick up. That puts the green icon there. He's going to map out an area of where the robot should pick up the object from. It just moves it around, and the robot figures out that was an area search. He didn't have to select that from a menu. And now he's going to go off and train the visual appearance of that object while we continue talking. So as we continue here, I want to tell you about what this is like in factories. These robots we're shipping every day. They go to factories around the country. This is Mildred. Mildred's a factory worker in Connecticut. She's worked on the line for over 20 years. One hour after she saw her first industrial robot, she had programmed it to do some tasks in the factory. She decided she really liked robots. And it was doing the simple repetitive tasks that she had had to do beforehand. Now she's got the robot doing it. When we first went out to talk to people in factories about how we could get robots to interact with them better, one of the questions we asked them was, "Do you want your children to work in a factory?" The universal answer was "No, I want a better job than that for my children." And as a result of that, Mildred is very typical of today's factory workers in the U.S. They're older, and they're getting older and older. There aren't many young people coming into factory work. And as their tasks become more onerous on them, we need to give them tools that they can collaborate with, so that they can be part of the solution, so that they can continue to work and we can continue to produce in the U.S. And so our vision is that Mildred who's the line worker becomes Mildred the robot trainer. She lifts her game, like the office workers of the 1980s lifted their game of what they could do. We're not giving them tools that they have to go and study for years and years in order to use. They're tools that they can just learn how to operate in a few minutes. There's two great forces that are both volitional but inevitable. That's climate change and demographics. Demographics is really going to change our world. This is the percentage of adults who are working age. And it's gone down slightly over the last 40 years. But over the next 40 years, it's going to change dramatically, even in China. The percentage of adults who are working age drops dramatically. And turned up the other way, the people who are retirement age goes up very, very fast, as the baby boomers get to retirement age. That means there will be more people with fewer social security dollars competing for services. But more than that, as we get older we get more frail and we can't do all the tasks we used to do. If we look at the statistics on the ages of caregivers, before our eyes those caregivers are getting older and older. That's happening statistically right now. And as the number of people who are older, above retirement age and getting older, as they increase, there will be less people to take care of them. And I think we're really going to have to have robots to help us. And I don't mean robots in terms of companions. I mean robots doing the things that we normally do for ourselves but get harder as we get older. Getting the groceries in from the car, up the stairs, into the kitchen. Or even, as we get very much older, driving our cars to go visit people. And I think robotics gives people a chance to have dignity as they get older by having control of the robotic solution. So they don't have to rely on people that are getting scarcer to help them. And so I really think that we're going to be spending more time with robots like Baxter and working with robots like Baxter in our daily lives. And that we will -- Here, Baxter, it's good. And that we will all come to rely on robots over the next 40 years as part of our everyday lives. Thanks very much. (Applause)
亞瑟·查理斯·克拉克 上世紀50年代著名的科幻小說家 曾說過:“從短期看來,我們高估了科技; 但從長期而言,我們卻低估了它” 隨著人工智能和機器人技術的發展 我們開始害怕某些工作將被取代 正是我們高估科技短期影響的一種代表 但我擔心的是從長遠看, 我們能否達到所需要的科技水平 人口的增長讓我們需要更多人手 我們的社會將不得不建立在這些鋼鐵機器的肩膀上。 所以,我擔心的是我們沒有足夠的機器人 科技會導致失業的想法其實由來已久 1975年,史賓塞·屈賽 和 凯瑟琳·赫本主演主演過一部電影 你知道最後最後結局如何嗎? 史賓塞·屈賽 弄來了一台電腦,一台1957年的大型機 幫助那些圖書管理員 公司的圖書管理員需要負責回答高官們的問題。例如, “聖誕老人的馴鹿叫什麼名字?” 圖書管理員們就回去把答案找出來。 這些大型計算機就會幫助他們 當然,一台1957年的大型機也不見得對這工作有多大幫助 然而圖書管理員們依舊害怕他們會失業 但事實上事情並非如此。 在1957年之後很長的一段時間裡, 圖書管理員的數量反而增長了 直到互聯網出現, 網絡出現,搜索引擎出現 對圖書管理員的需求才開始下降。 同時,我認為在1957年所有人都完完全全低估了 我們今天握在手中以及裝在口袋中的這些東西的科技含量 只需一瞬間,我們就可以知道聖誕老人的馴鹿的名字, 抑或是任何我們想問的 順帶一提,圖書管理員的工資增速 曾在一段時間內高過了全美其他崗位的工資水平, 因為圖書管理員成為了電腦的同夥 電腦成為了他們的工具, 同時他們也獲取了更多其他可用的工具 讓效率變得更高。 同樣的事情也發生在辦公室裡 以前,人們處理報表的方式是 把數據寫在許多不同的紙張 一一用手計算。 但是有趣的事情發生了。 隨著1980年的電腦革命, 空白表格程式沒有取代辦公族, 反而受到他們的青睞 辦公族變身成為程式設計師, 當他們成為空白表格的程式設計師 他們的工作更有效率了。 他們不用再做那些繁瑣的計算, 他們可以做更多其他工作。 今天,我們在日常生活中也能見到機器人的身影。 左邊是一台 iRobot 公司產的軍用機械人 PackBot 當士兵們穿越伊拉克和阿富汗戰場的雷區時, 他們不再像 2002 年之前那樣, 穿著防彈背心拿著探棒到處戳, 現在他們派機器人去 讓機器人負責這些危險的工作 在右邊是匹玆堡的一家名為 Aethon 的公司 生產的 TUG 機器人。 全美近百家醫院正在使用這些機器人 它們把床單送去洗衣房。 把髒盤子送回廚房 從藥房取藥送給病人 這使得護士和他們的助手 從那些到處搬東西的機械化勞動中解放, 花更多的時間的陪患者。 事實上,機器人已經普及在我們生活的很多層次。 但是如果談及工業機器人,人們可能還是會有些害怕的, 因為工業機器人有可能會傷及周圍的人。 如果要為它們設計程序,你需要理解六維向量和四元空間。 一般人無法和它們溝通。 我認為一旦科技完全取代了原本的工人 這樣的科技就有問題了 我們確實需要思考一下如何讓工人 可以和這些高科技產物相互合作。 所以今天我想聊聊我們曾經談到過的 Baxter 機器人。 Baxter 在我看來是第一批 通過一些工業設定就可以和普通人互相溝通的機器人 讓我們來看看 Baxter。 這位是 Rethink Robotics 的克里斯·哈伯特 在這裡我們有一個輸送帶 如果亮度不是過高的話 對了,對了。Baxter 從輸送帶上拿起了零件。 接著它把零件拿過來放下。 然後再回去取下一個零件。 有趣的是,Baxter 也具備一些基本的常識。 順帶一提,它的眼睛去哪兒了? 眼睛在那邊的螢幕上。 它會看著機器人要移動的方向。 因此和機器人一起工作的人 可以明白機器人要移向哪裡 而不會被他的動向嚇到。 現在克里斯從它手裡拿走一個零件, 這時 Baxter 不會繼續嘗試將那零件移過去放下; 它會返回原位,因為它意識到自己要去取下一個零件。 在拿取和移動零件上Baxter已有了一些常識。 同時與 Baxter 一起工作也是很安全的。 你也許不會想和現在市面上的工業機器人一起工作。 但是和Baxter一起是安全的 它能夠感覺阻力,從而明白克里斯在那裡。 它不會推他導致傷到他 但是我認為 Baxter 最有意思的還是它的用戶界面。 現在克里斯要過去抓住它另一只手臂 當他抓住一只手的時候, Baxter 就進入了無動力重力補償模式, 同時這樣的圖像出現在螢幕上 你可以看到一些圖標出現在螢幕的右邊, 它們代表了 Baxter 的右臂。 他打算把那些東西放到這裡來, 按下一個按鈕,然後讓它放下手裡的東西。 然後機器人明白了,“嗯,他一定是要我把這個東西放下” 它在需要放零件的地方標了個圖標。 它把機器手移到這裡,併起它的手指, 機器人明白克里斯要它撿起一個零件 在那邊標一個綠色的圖標。 克里斯現在要劃出一塊區域, 讓機器人從這塊區域裡取零件。 他只是把機械手臂到處移動, 機器人就明白這是一塊搜索區域。 他不用在選單中選擇。 他現在要離開一會兒,去教會機器人識別零件。 現在我們繼續聊。 說到這裡, 我先要告訴你們這些機器人在工廠裡是怎麼工作的。 這些每天運出的這些機器人, 被送往遍佈全美的工廠。 這位是米爾德里德。 米爾德里德是康涅狄格的一名工人。 她在生產線上工作了20多年。 就在她見到她生平的第一個工業機器人的一個小時以後, 她就已經教會了這台機器人一些工廠裡的工作。 她確實非常喜歡機器人。 機器人正在做那些她之前不得不做的重複性工作。 現在機器人代替她做這些。 在我們最開始走到工廠裡與那裡的人們談論 我們如何更好的讓機器人和他們合作時, 我們問的其中一個問題是, “你想讓你的孩子在工廠工作嗎?” 所有答案都是,“不,我想我孩子有個更好的工作。” 其結果是,米爾德里德就是現在美國一個很典型的 工廠工人。 他們都比較年長,並在不斷走向衰老。 很少有年輕人願意在工廠工作。 隨著他們肩負的工作變得日益繁重, 我們需要提供他們一些可以幫助他們的工具, 使他們可以成為解決方案的一部分, 使他們可以繼續留在工作崗位上, 也是美國的製造業得以持續。 所以我們期望米爾德里德可以從一個流水線工人 轉變為一個機器人教練。 她改變了她的工作性質, 就如同上世紀 80 年代的辦公室一族一樣 我們不會提供他們那些需要花好幾年才能學會使用的工具。 我們提供的工具只需幾分鐘就可以學會操作。 這世界上有兩種必須出現、無法避免的力量 那就是氣候變遷和人口變化 人口的轉變將確確實實的改變我們的世界。 這是處於工作年齡的成年人佔整體成年人數的百分比。 在過去的40年中輕微的下跌 但是在未來的40年,它將有顯著的變化,即便是在中國。 處於工作年齡的成年人比例將顯著下降。 另一方面,隨著嬰兒潮一代逐步步入退休年齡, 處於退休年齡的人將越來越多。 那意味著將有更多的人需要服務 社會福利的資金卻會減少 不止如此,隨著年齡的增長,我們將變得更加脆弱 以至於我們沒辦法完成那些我們曾經可以做到的事情。 如果我們看一下社工的年齡統計數據, 我們所看到的是這些社工正變得越來越年長。 而統計結果也正表明了這一點。 隨著那些越發年邁的退休者的數量的增加, 能夠照顧他們的人缺日趨減少。 所以我們真切的感受到 我們不得不讓機器人去幫助他們。 我並不是在說機器人伴侶。 我指的是有機器人來做一些 一般我們可以自己完成 但隨著年齡增長變得艱難的日常瑣事。 例如將食物從車裡搬出來,上樓搬進廚房。 或者,等我們再老一點, 開著車去見朋友。 我認為通過控制機器人解決問題 那些年邁的人將獲得更多尊嚴。 因此他們不用在依靠那些日漸稀缺的人們去幫助他們。 我相信我們將與 Baxter 這樣的機器人 一起度過更多的時間 並在日常生活中與像 Baxter 這樣的機器人合作。 看,Baxter,它很不錯。 在接下來的40年中 我們都會需要依賴機器人 它將成為我們日常生活的一部分 謝謝各位 (掌聲)