PBS高端访谈:新的科学研究发现:你的工作将会被外包给机器人?

 　　GWEN IFILL: Have you ever worried you might lose your job to a robot? I have.

　　Hari Sreenivasan finds it could well happen with advances in artificial intelligence, or A.I., transforming the work force.

　　That’s the latest report in our series on invention and innovation, Breakthroughs.

　　MAN: Oh, all in?

　　HARI SREENIVASAN: In a closely watched brains vs. artificial intelligence poker¹ match held in Pittsburgh earlier this month, humans pulled off a slim win over a computer program called Claudico.

　　MAN: All right. Good job.

　　MAN: Good game, guys. Good game.

　　MAN: Good game.

　　HARI SREENIVASAN: Tuomas Sandholm, a computer scientist at Carnegie Mellon University, created the algorithms that run Claudico’s A.I.

　　TUOMAS SANDHOLM, Carnegie Mellon University: Those algorithms figure out how you should act strategically, how do you avoid or deal with humans trying to deceive you, and how do you deceive humans?

　　HARI SREENIVASAN: Sandholm predicts Claudico will be able to beat its human opponents within one to five years, much to the chagrin³ of Bjorn Li, the leading poker player in this tournament.

　　BJORN LI: When that happens, poker will pretty much be dead.

　　HARI SREENIVASAN: But putting pro² poker players out of work is not what Sandholm focuses all his time on. There are other things that Claudico can already do better than humans.

　　TUOMAS SANDHOLM: In my lab, we have developed an algorithm for solving the matching problem for the nationwide kidney exchange for 60 percent of the transplant centers in the U.S. And there, twice a week, our algorithms make the transplantation plan for the whole country without any manual intervention⁴. When there is scarcity⁵ of organs, the A.I. is making those decisions in an optimal⁶ way.

　　HARI SREENIVASAN: Matching the right kidney to the right patient is one example of an algorithmic artificial intelligence. But there are much larger demonstrations⁷ hitting the road, quite literally⁸.

　　Daimler has developed a prototype dubbed⁹ the Freightliner Inspiration Truck that’s being test-driven across Nevada. The hope is that computer-driven trucks can reduce the number of accidents. There are currently 5,000 fatalities¹⁰ a year involving trucks. Drivers would function more like pilots, overseeing computerized systems.

　　But it begs the question: What jobs will survive in a new economy driven by automation?

　　Remember Ken¹¹ Jennings, the “Jeopardy” game show champion who lost to IBM’s Watson in 2011? He says the writing is on the wall. Here he is in a TEDx talk.

　　KEN JENNINGS: And I remember standing¹² there behind the podium, as I could hear that little insectoid thumb. And you could hear that little tick, tick, tick, tick, tick, tick.

　　KEN JENNINGS: And I remember thinking, you know, this is it. I felt obsolete¹³. I felt like a Detroit factory worker of the ’80s seeing a robot that could now do his job on the assembly line. And it was frigging demoralizing.

　　HARI SREENIVASAN: It’s not just quiz show contestants¹⁴ that are at risk. As more and more jobs are automated¹⁵, Jennings’ experience could be a harbinger of things to come for American workers.

　　That’s the argument made in a new book, “Rise of the Robots: Technology and the Threat of a Jobless Future,” by Martin Ford¹⁶.

　　MARTIN FORD, Author, “Rise of the Robots”: Going forward, we may see automation kind of unfold in a top-heavy pattern, where a lot of the best jobs are the ones to get impacted. Lawyers, pharmacists, certain areas of medicine like pathology and radiology, any kind of white-collar job where you are sitting at a computer at a desk, well, the people who you might call office drones, those are going to be very susceptible¹⁷ to this.

　　HARI SREENIVASAN: And there could be major disruptions to the U.S. economy, says Daphne Koller. She’s an A.I. scientist, and also president of the massive online learning company Coursera.

　　DAPHNE KOLLER, Coursera: We are already starting to see jobs that were thought of as intelligent being outsourced to computers.

　　So, for example, a large part of a paralegal’s job, which is hunting down the relevant references for a particular problem, is something that you would have thought requires intelligence. And now there are pretty good software systems that do not 100 percent of a paralegal’s job, but 80 to 90 percent.

　　HARI SREENIVASAN: Will artificial intelligence software do to the paralegal what the tractor did to the farmer?

　　DAPHNE KOLLER: It is quite likely that that will happen. And I think that there will be entire job categories that will go away.

　　HARI SREENIVASAN: We humans have always been resilient. With each industrial revolution, we have adapted, creating new jobs with new technologies.

　　DAPHNE KOLLER: The optimistic perspective is that this will happen here, and that the jobs that will be created will by nature be higher and more cognitively¹⁸ interesting jobs that are beyond the spectrum¹⁹ of what an artificial intelligence program can do.

　　HARI SREENIVASAN: Leaving the less interesting jobs to robotic helpers like Botlr, an automated bellhop who cruises the halls of this Aloft Hotel. Is that such a bad thing?

　　Stuart Russell, who directs the A.I. lab at the University of California at Berkeley, doesn’t think so.

　　STUART RUSSELL, University of California, Berkeley: Some people think that, inevitably²⁰, every robot that does any task is a bad thing for the human race, because it could be taking a job away.

　　But that isn’t necessarily true. You can also think of the robot as making a person more productive and enabling people to do things that are currently economically infeasible. But a person plus a robot or a fleet of robots could do things that would be really useful.

　　HARI SREENIVASAN: A perhaps simple example, cleaning up graffiti.

　　STUART RUSSELL: In many, many cities, the graffiti is just left because it’s too expensive. But if I had a team of robots that I could take around the city with me and point them to what needed to be cleaned up, I could get 10 times as much done. And there will be positions for graffiti-cleaning supervisors²¹, which didn’t exist before.

　　HARI SREENIVASAN: Graffiti-cleaning supervising robots might exist in the future, but our economy is already evolving. There are plenty of jobs that didn’t exist 10 years ago that are now in high demand in fields like digital marketing²² and data analysis.

　　In fact, according to McKinsey & Company, the United States faces a shortage of data analysts²³. Almost 190,000 people are needed to analyze²⁴ and understand big data. But will those jobs ultimately be filled by people or by deep learning machines?

　　Deep learning is a new type of A.I. that relies on neural²⁵ networks. They’re computer programs modeled after the human brain and nervous system.

　　MAN: Hey, guys. How’s the training page looking?

　　HARI SREENIVASAN: At the Palo Alto office of MetaMind, engineers are using the technology to help computers see by quickly identifying images and placing them in categories.

　　The software can also understand nuance²⁶ in the written word.

　　Richard Socher is co-founder and CTO. He says the technology will aid humans, not replace them.

　　RICHARD SOCHER, MetaMind: If you can bring the intelligence of the smartest people in a field, instill it in an algorithm with deep learning, you could really help a lot of people.

　　HARI SREENIVASAN: One example, he says, is in the field of medicine.

　　RICHARD SOCHER: If the best doctors in the world train an algorithm to find various different problems in C.T. scans or in X-rays, mammograms, for instance, you could build an algorithm that is almost as good as the best doctors in the world.

　　A human can only look at so many mammograms in their lifetime. An algorithm could look at millions and millions, and eventually find subtle things that may have not even been that obvious to the human eye.

　　HARI SREENIVASAN: So, how will society adapt to a computer intelligence that can do work which, until now, only humans could?

　　DAPHNE KOLLER: What people have going for them that computers as of yet don’t is the incredible adaptability²⁷ of the human mind, the ability to learn new skills, the ability to really adapt to unexpected situations.

　　And so what we really need to do is to help people become even better at that.

　　HARI SREENIVASAN: Just like in a poker game, we don’t know what the outcome will be. We humans are raising the stakes as we continue to drive advances in A.I. technology. So, it will be up to us to stay at the table.

　　For the PBS NewsHour, I’m Hari Sreenivasan.

　　JUDY WOODRUFF: And you can watch more stories from our Thinking Machines series on our Web site, PBS.org/NewsHour.

1 poker   

n.扑克；vt.烙制

参考例句：

• He was cleared out in the poker game.他打扑克牌，把钱都输光了。
• I'm old enough to play poker and do something with it.我打扑克是老手了，可以玩些花样。

2 pro   

n.赞成，赞成的意见，赞成者

参考例句：

• The two debating teams argued the question pro and con.辩论的两组从赞成与反对两方面辩这一问题。
• Are you pro or con nuclear disarmament?你是赞成还是反对核裁军？

3 chagrin   

n.懊恼；气愤；委屈

参考例句：

• His increasingly visible chagrin sets up a vicious circle.他的明显的不满引起了一种恶性循环。
• Much to his chagrin,he did not win the race.使他大为懊恼的是他赛跑没获胜。

4 intervention   

n.介入，干涉，干预

参考例句：

• The government's intervention in this dispute will not help.政府对这场争论的干预不会起作用。
• Many people felt he would be hostile to the idea of foreign intervention.许多人觉得他会反对外来干预。

5 scarcity   

n.缺乏，不足，萧条

参考例句：

• The scarcity of skilled workers is worrying the government.熟练工人的缺乏困扰着政府。
• The scarcity of fruit was caused by the drought.水果供不应求是由于干旱造成的。

6 optimal   

adj.最适宜的；最理想的；最令人满意的

参考例句：

• What is the optimal mix of private and public property rights in natural resources?私人和国家的自然资源产权的最适宜的组合是什么？
• Optimal path planning is a key link for the sailing contest.帆船最优行驶路径规划是帆船比赛取胜的关键环节。

7 demonstrations   

证明( demonstration的名词复数 )； 表明； 表达； 游行示威

参考例句：

• Lectures will be interspersed with practical demonstrations. 讲课中将不时插入实际示范。
• The new military government has banned strikes and demonstrations. 新的军人政府禁止罢工和示威活动。

8 literally   

adv.照字面意义，逐字地；确实

参考例句：

• He translated the passage literally.他逐字逐句地翻译这段文字。
• Sometimes she would not sit down till she was literally faint.有时候，她不走到真正要昏厥了，决不肯坐下来。

9 dubbed   

v.给…起绰号( dub的过去式和过去分词 )；把…称为；配音；复制

参考例句：

• Mathematics was once dubbed the handmaiden of the sciences. 数学曾一度被视为各门科学的基础。
• Is the movie dubbed or does it have subtitles? 这部电影是配音的还是打字幕的？ 来自《简明英汉词典》

10 fatalities   

n.恶性事故( fatality的名词复数 )；死亡；致命性；命运

参考例句：

• Several people were injured, but there were no fatalities. 有几个人受伤，但没有人死亡。
• The accident resulted in fatalities. 那宗意外道致多人死亡。 来自《简明英汉词典》

11 ken   

n.视野，知识领域

参考例句：

• Such things are beyond my ken.我可不懂这些事。
• Abstract words are beyond the ken of children.抽象的言辞超出小孩所理解的范围.

12 standing   

n.持续，地位；adj.永久的，不动的，直立的，不流动的

参考例句：

• After the earthquake only a few houses were left standing.地震过后只有几幢房屋还立着。
• They're standing out against any change in the law.他们坚决反对对法律做任何修改。

13 obsolete   

adj.已废弃的，过时的

参考例句：

• These goods are obsolete and will not fetch much on the market.这些货品过时了，在市场上卖不了高价。
• They tried to hammer obsolete ideas into the young people's heads.他们竭力把陈旧思想灌输给青年。

14 contestants   

n.竞争者，参赛者( contestant的名词复数 )

参考例句：

• The competition attracted over 500 contestants representing 8 different countries. 这次比赛吸引了代表8个不同国家的500多名参赛者。
• Two candidates are emerging as contestants for the presidency. 两位候选人最终成为总统职位竞争者。 来自《简明英汉词典》

15 automated   

a.自动化的

参考例句：

• The entire manufacturing process has been automated. 整个生产过程已自动化。
• Automated Highway System (AHS) is recently regarded as one subsystem of Intelligent Transport System (ITS). 近年来自动公路系统(Automated Highway System,AHS)，作为智能运输系统的子系统之一越来越受到重视。

16 Ford   

n.浅滩，水浅可涉处；v.涉水，涉过

参考例句：

• They were guarding the bridge,so we forded the river.他们驻守在那座桥上，所以我们只能涉水过河。
• If you decide to ford a stream,be extremely careful.如果已决定要涉过小溪，必须极度小心。

17 susceptible   

adj.过敏的，敏感的；易动感情的，易受感动的

参考例句：

• Children are more susceptible than adults.孩子比成人易受感动。
• We are all susceptible to advertising.我们都易受广告的影响。

18 cognitively   

参考例句：

• Cognitively,man,the subject of cognition,must classify and categorize the objects. 从认知学角度来看 ，作为认知主体的人对于认知对象必须进行分类和范畴化。 来自互联网
• Cognitively, reference can be studied along with information processing of human mind. 从认知的角度看，要研究人类思维的信息处理过程。 来自互联网

19 spectrum   

n.谱，光谱，频谱；范围，幅度，系列

参考例句：

• This is a kind of atomic spectrum.这是一种原子光谱。
• We have known much of the constitution of the solar spectrum.关于太阳光谱的构成，我们已了解不少。

20 inevitably   

adv.不可避免地；必然发生地

参考例句：

• In the way you go on,you are inevitably coming apart.照你们这样下去，毫无疑问是会散伙的。
• Technological changes will inevitably lead to unemployment.技术变革必然会导致失业。

21 supervisors   

n.监督者，管理者( supervisor的名词复数 )

参考例句：

• I think the best technical people make the best supervisors. 我认为最好的技术人员可以成为最好的管理人员。 来自辞典例句
• Even the foremen or first-level supervisors have a staffing responsibility. 甚至领班或第一线的监督人员也有任用的责任。 来自辞典例句

22 marketing   

n.行销，在市场的买卖，买东西

参考例句：

• They are developing marketing network.他们正在发展销售网络。
• He often goes marketing.他经常去市场做生意。

23 analysts   

分析家，化验员( analyst的名词复数 )

参考例句：

• City analysts forecast huge profits this year. 伦敦金融分析家预测今年的利润非常丰厚。
• I was impressed by the high calibre of the researchers and analysts. 研究人员和分析人员的高素质给我留下了深刻印象。

24 analyze   

vt.分析，解析 (=analyse)

参考例句：

• We should analyze the cause and effect of this event.我们应该分析这场事变的因果。
• The teacher tried to analyze the cause of our failure.老师设法分析我们失败的原因。

25 neural   

adj.神经的，神经系统的

参考例句：

• The neural network can preferably solve the non- linear problem.利用神经网络建模可以较好地解决非线性问题。
• The information transmission in neural system depends on neurotransmitters.信息传递的神经途径有赖于神经递质。

26 nuance   

n.(意义、意见、颜色)细微差别

参考例句：

• These users will easily learn each nuance of the applications they use.这些用户会很快了解他们所使用程序的每一细微差别。
• I wish I hadn't become so conscious of every little nuance.我希望我不要变得这样去思索一切琐碎之事。

27 adaptability   

n.适应性

参考例句：

• It has a wide range of adaptability.它的应用性广。