Shyam Sankar: The rise of human-computer cooperation

Translator: Joseph Geni Reviewer: Morton Bast

翻译人员: Yi Zong 校对人员: Chunting Guo

I'd like to tell you about two games of chess.

我想告诉你们两场象棋比赛。

The first happened in 1997, in which Garry Kasparov,

一场发生在1997年，卡斯帕罗夫，

a human, lost to Deep Blue, a machine.

一个人类， 输给了‘深蓝’，一部机器。

To many, this was the dawn of a new era,

对许多人来说，这是一个新时代的黎明，

one where man would be dominated by machine.

一个人被机器统治的时代。

But here we are, 20 years on, and the greatest change

但现在的我们，20年已经过去了，而最能改变

in how we relate to computers is the iPad,

我们与电脑之间关系的是IPAD，

not HAL.

不是 HAL。

The second game was a freestyle chess tournament

第二场是自由式国际象棋锦标赛

in 2005, in which man and machine could enter together

在2005年，人类与机器可以一起进入比赛

as partners, rather than adversaries, if they so chose.

以合作伙伴的身份，而不是敌人，如果他们这样选择。

At first, the results were predictable.

起初，结果是可以预测的。

Even a supercomputer was beaten by a grandmaster

即使是一台超级计算机也会输给特级大师

with a relatively weak laptop.

和一台相对较弱的便携式计算机。

The surprise came at the end. Who won?

可结局令人惊讶。谁赢了？

Not a grandmaster with a supercomputer,

不是使用超级计算机的大师，

but actually two American amateurs

而实际上是两个美国业余选手

using three relatively weak laptops.

和他们使用的三台相对较弱的笔记本电脑。

Their ability to coach and manipulate their computers

他们有能力知道和操纵他们的计算机

to deeply explore specific positions

从而深入探索具体的位置

effectively counteracted the superior chess knowledge

以有效的方法抵消

of the grandmasters and the superior computational power

大师和卓越计算的优越的国际象棋知识

of other adversaries.

和其他对手。

This is an astonishing result: average men,

这是一个令人吃惊的结果： 普通男性，

average machines beating the best man, the best machine.

一般的计算机击败最好的人和最好的机器。

And anyways, isn't it supposed to be man versus machine?

不管怎么说，不应该是机器于人对战吗？

Instead, it's about cooperation, and the right type of cooperation.

相反，它是关于合作和正确的合作方式。

We've been paying a lot of attention to Marvin Minsky's

近 50 年，我们一直集中大量的精力到 Marvin Minsky

vision for artificial intelligence over the last 50 years.

的人工智能的愿景。

It's a sexy vision, for sure. Many have embraced it.

它是一个性感的远景，这是肯定的。很多人已经接受它了。

It's become the dominant school of thought in computer science.

它已成为计算机科学的主流学派。

But as we enter the era of big data, of network systems,

但是，当我们进入了大数据的时代、 网络系统、

of open platforms, and embedded technology,

开放平台和嵌入式技术，

I'd like to suggest it's time to reevaluate an alternative vision

我想建议是重新评估另一个的远景的时候了

that was actually developed around the same time.

这实际上是大约在同一时间进行开发的。

I'm talking about J.C.R. Licklider's human-computer symbiosis,

我讨论的是 J.C.R.Licklider 的人机共生，

perhaps better termed "intelligence augmentation," I.A.

或许更好地被称为"智能强化"一I.A.

Licklider was a computer science titan who had a profound

Licklider 是一位计算机科学巨人

effect on the development of technology and the Internet.

他对技术和互联网发展有非常深刻的影响。

His vision was to enable man and machine to cooperate

他的设想是，使人与机器进行合作

in making decisions, controlling complex situations

从而作出决定，控制复杂的情况

without the inflexible dependence

而不是死板的依赖

on predetermined programs.

于预先设定的程序。

Note that word "cooperate."

请注意，这个词语"合作"。

Licklider encourages us not to take a toaster

Licklider 鼓励我们不是用一个烤面包机

and make it Data from "Star Trek,"

并使其变成《 星际迷航 》中的科技，

but to take a human and make her more capable.

而要采取一个人，并使她更有能力。

Humans are so amazing -- how we think,

人类如此惊人 — 我们的思维

our non-linear approaches, our creativity,

我们的非线形方法，我们的创造力，

iterative hypotheses, all very difficult if possible at all

迭代的假设，都很难

for computers to do.

让计算机做到类似的事。

Licklider intuitively realized this, contemplating humans

Licklider 直观地认识了这一点，考虑人类

setting the goals, formulating the hypotheses,

设定目标，提出假说，

determining the criteria, and performing the evaluation.

确定的标准，并进行评价。

Of course, in other ways, humans are so limited.

当然，在其他方面，人类是如此有限。

We're terrible at scale, computation and volume.

我们在大规模、 计算和容量方面做得很遭。

We require high-end talent management

我们需要高端的人才管理

to keep the rock band together and playing.

以保持摇滚乐队一起演奏。

Licklider foresaw computers doing all the routinizable work

Licklider 预见到所有的程序化的工作可以由计算机完成

that was required to prepare the way for insights and decision making.

这需要预先准备目标和决策的方法。

Silently, without much fanfare,

安静得，没有大张旗鼓，

this approach has been compiling victories beyond chess.

这种做法已经超越了象棋的胜利。

Protein folding, a topic that shares the incredible expansiveness of chess —

蛋白质排列，一个同样令人难以置信的广阔的国际象棋的话题 — —

there are more ways of folding a protein than there are atoms in the universe.

蛋白质排列方式要比在宇宙中的原子更多。

This is a world-changing problem with huge implications

这对改变世界问题启示了

for our ability to understand and treat disease.

我们有能力了解和治疗疾病。

And for this task, supercomputer field brute force simply isn't enough.

而对于这个任务，只有超级计算机的蛮力还不够。

Foldit, a game created by computer scientists,

Foldit，计算机科学家创建的一个游戏，

illustrates the value of the approach.

说明了这个方法的价值。

Non-technical, non-biologist amateurs play a video game

非技术性、 非生物学家业余玩的视频游戏

in which they visually rearrange the structure of the protein,

在其中他们直观地重新排列蛋白质的结构，

allowing the computer to manage the atomic forces

允许此计算机管理原子的力量

and interactions and identify structural issues.

和互动，并识别结构问题。

This approach beat supercomputers 50 percent of the time

这种方法以50%的几率击败了超级计算机

and tied 30 percent of the time.

以30%的几率战平。

Foldit recently made a notable and major scientific discovery

Foldit最近取得一个显著并重大的科学发现

by deciphering the structure of the Mason-Pfizer monkey virus.

它破译梅森辉瑞猴病毒的结构。

A protease that had eluded determination for over 10 years

一种躲避测定十多年的蛋白酶

was solved was by three players in a matter of days,

被三名球员在仅仅几天时间就解决了，

perhaps the first major scientific advance

也许这是第一次重大科学进展

to come from playing a video game.

源于玩视频游戏。

Last year, on the site of the Twin Towers,

去年，该站点的双子塔，

the 9/11 memorial opened.

9/11 纪念馆开幕。

It displays the names of the thousands of victims

它显示了数千名受害者的名称

using a beautiful concept called "meaningful adjacency."

通过一个美丽的概念称为"有意义的邻接"。

It places the names next to each other based on their

它把他们的名字彼此相邻的安排在一起，根据

relationships to one another: friends, families, coworkers.

从一个到另一个人的关系： 朋友、 家人、 同事。

When you put it all together, it's quite a computational

当你把它放在一起时，这是相当大的计算

challenge: 3,500 victims, 1,800 adjacency requests,

挑战：3,500名 受害者、 1,800名邻接请求，

the importance of the overall physical specifications

整体物理属性的重要性

and the final aesthetics.

和最后的审美。

When first reported by the media, full credit for such a feat

当第一次被媒体报道时，这件壮举完全归功

was given to an algorithm from the New York City

给了纽约

design firm Local Projects. The truth is a bit more nuanced.

本地的设计公司设计的运算法则。事实真相更为微妙。

While an algorithm was used to develop the underlying framework,

虽然一种运算法则被用来开发基本框架，

humans used that framework to design the final result.

人类利用这一框架来设计最终的结果。

So in this case, a computer had evaluated millions

所以在这种情况下，计算机已评估了数百万种

of possible layouts, managed a complex relational system,

可能的布局， 管理一个复杂的关系系统，

and kept track of a very large set of measurements

和跟踪大量测量数据

and variables, allowing the humans to focus

和变量，使人类能够专注于

on design and compositional choices.

设计和组合的选择。

So the more you look around you,

所以你越常环顾你的周围，

the more you see Licklider's vision everywhere.

在各个地方，您越常看到 Licklider 的愿景。

Whether it's augmented reality in your iPhone or GPS in your car,

无论是已经现实在你的iPhone的技术或在你车上的GPS的技术，

human-computer symbiosis is making us more capable.

人机共生使我们更有能力。

So if you want to improve human-computer symbiosis,

因此，如果您想要改善人机共生

what can you do?

你可以做什么？

You can start by designing the human into the process.

您可以从将人设计到过程中开始。

Instead of thinking about what a computer will do to solve the problem,

而不是思考如何让计算机解决问题，

design the solution around what the human will do as well.

围绕着人去设计结局方案。

When you do this, you'll quickly realize that you spent

当您执行此操作时，您很快就会发现你花了

all of your time on the interface between man and machine,

你所有的时间上在人和机器之间的接口上，

specifically on designing away the friction in the interaction.

特别是关于设计互动中的摩擦。

In fact, this friction is more important than the power

事实上，这种冲突比

of the man or the power of the machine

人或机器的力量更重要，

in determining overall capability.

从整体能力上讲。

That's why two amateurs with a few laptops

这就是为什么几个笔记本电脑和两个业余选手

handily beat a supercomputer and a grandmaster.

轻松击败了一台超级计算机和特级大师。

What Kasparov calls process is a byproduct of friction.

卡斯帕罗夫称这个过程是摩擦的副产品。

The better the process, the less the friction.

越好的过程，摩擦越少。

And minimizing friction turns out to be the decisive variable.

尽量减少摩擦原来是决定性变量。

Or take another example: big data.

或再举一个例子： 海量数据。

Every interaction we have in the world is recorded

我们在世界上每个互动都被记录着

by an ever growing array of sensors: your phone,

由与日俱增的传感器：您的电话，

your credit card, your computer. The result is big data,

您的信用卡，您的计算机。其结果是大量的数据，

and it actually presents us with an opportunity

同时它实际上提供了我们一个机会

to more deeply understand the human condition.

去更深入地理解人类的特点。

The major emphasis of most approaches to big data

处理海量数据的主要方法

focus on, "How do I store this data? How do I search

是集中在，"如何存储这些数据？如何搜索

this data? How do I process this data?"

这些数据？如何处理这些数据？"

These are necessary but insufficient questions.

这些都是必要但不完全的问题。

The imperative is not to figure out how to compute,

当务之急是不弄清楚如何计算，

but what to compute. How do you impose human intuition

但用什么来计算。如何加入的人类直觉

on data at this scale?

在这种规模的数据上？

Again, we start by designing the human into the process.

又一次，我们开始设计把人类引入这一进程。

When PayPal was first starting as a business, their biggest

PayPal 作为一家企业，当他们第一次启动时，他们最大的

challenge was not, "How do I send money back and forth online?"

挑战不是，"如何在线转账？"

It was, "How do I do that without being defrauded by organized crime?"

而是，"如何避免有组织犯罪的诈骗？"

Why so challenging? Because while computers can learn

为什么如此具有挑战性？因为虽然计算机能学到

to detect and identify fraud based on patterns,

探测和识别基于模式的欺诈，

they can't learn to do that based on patterns

他们学不会做基于模式

they've never seen before, and organized crime

之外的判断，这同有组织犯罪

has a lot in common with this audience: brilliant people,

有很多共同点： 聪明，

relentlessly resourceful, entrepreneurial spirit — (Laughter) —

足智多谋、 有创业精神 —（笑声）—

and one huge and important difference: purpose.

还有一个重大的区别：目的。

And so while computers alone can catch all but the cleverest

所以在单独的计算机可以捕获所有得同时，最聪明

fraudsters, catching the cleverest is the difference

的诈骗犯捕捉最聪明的，区别就是

between success and failure.

成功与失败。

There's a whole class of problems like this, ones with

像这类的问题有很多，都是

adaptive adversaries. They rarely if ever present with a

相互适应。他们很少显示出

repeatable pattern that's discernable to computers.

可以辨认到计算机的可重复执行的模式。

Instead, there's some inherent component of innovation or disruption,

相反，有一些继承下来的创新或中断的部分

and increasingly these problems are buried in big data.

同时这些问题越来越多地被藏在了大量的数据中。

For example, terrorism. Terrorists are always adapting

例如，恐怖主义。恐怖分子总可以适应这种

in minor and major ways to new circumstances, and despite

次要和主要方式的新环境，而且即使

what you might see on TV, these adaptations,

在电视上，你可能会看到这些适应能力，

and the detection of them, are fundamentally human.

和对他们的检测，基本上都是人类。

Computers don't detect novel patterns and new behaviors,

计算机不会检测新的模式和新的行为，

but humans do. Humans, using technology, testing hypotheses,

但人类可以。人类，利用技术、 测试假设，

searching for insight by asking machines to do things for them.

通过机器为他们寻找目标。

Osama bin Laden was not caught by artificial intelligence.

本 · 拉登不是被人工智能抓住的。

He was caught by dedicated, resourceful, brilliant people

他被抓住是因为专注、足智多谋和聪明的

in partnerships with various technologies.

人与各种技术的合作。

As appealing as it might sound, you cannot algorithmically

这听起来颇具吸引力，你不能通过计算

data mine your way to the answer.

数据的方式来挖掘你的答案。

There is no "Find Terrorist" button, and the more data

没有"找到恐怖分子"的按钮，同时越多的数据

we integrate from a vast variety of sources

，整合的来源越多，

across a wide variety of data formats from very

格式的种类越广，形成了一个非常

disparate systems, the less effective data mining can be.

迥异的系统，数据挖掘也更加低效。

Instead, people will have to look at data

相反，人们还要参考数据

and search for insight, and as Licklider foresaw long ago,

和搜索目标，Licklider 很久以前预见到的，

the key to great results here is the right type of cooperation,

成功的关键是正确的合作，

and as Kasparov realized,

正如卡斯帕罗夫意识到的，

that means minimizing friction at the interface.

这意味着，尽量减少操作界面的摩擦。

Now this approach makes possible things like combing

现在这种方法使得梳理

through all available data from very different sources,

所有可用且来源非常不同的数据成为了可能，

identifying key relationships and putting them in one place,

确定关键的关系，并将它们放在一个地方，

something that's been nearly impossible to do before.

之前看似几乎不可能完成的东西。

To some, this has terrifying privacy and civil liberties

对某些人来说，这会影响隐私和公民自由

implications. To others it foretells of an era of greater

的实行。对其他人来说，这预示了的一个更加伟大的时代

privacy and civil liberties protections,

对于保护隐私和公民自由，

but privacy and civil liberties are of fundamental importance.

但隐私和公民自由是根本也是最重要的。

That must be acknowledged, and they can't be swept aside,

必须承认，他们不能被抛在一边，

even with the best of intents.

即使是出于好意。

So let's explore, through a couple of examples, the impact

让我们探讨，通过几个例子，

that technologies built to drive human-computer symbiosis

科技构建驱动人机共生

have had in recent time.

最近产生的影响。

In October, 2007, U.S. and coalition forces raided

2007 年 10 月，美国和盟军部队突袭了

an al Qaeda safe house in the city of Sinjar

一个在辛贾尔市的基地组织的安全屋

on the Syrian border of Iraq.

位于叙利亚的伊拉克边界。

They found a treasure trove of documents:

他们发现一份宝贵的文档：

700 biographical sketches of foreign fighters.

700个外国战士的小传草稿。

These foreign fighters had left their families in the Gulf,

这些外国战士在海湾地区，他们离开他们

the Levant and North Africa to join al Qaeda in Iraq.

南欧和北非的家，加入在伊拉克境内的基地组织。

These records were human resource forms.

这些记录是人力资源管理的形式。

The foreign fighters filled them out as they joined the organization.

外国战士填写这些之后加入该组织。

It turns out that al Qaeda, too,

事实证明，基地组织，也，

is not without its bureaucracy. (Laughter)

并不是没有其官僚作风。（笑声）

They answered questions like, "Who recruited you?"

他们回答类似的问题，"谁聘请的你？"

"What's your hometown?" "What occupation do you seek?"

"你的家乡是哪儿？""你要求从事什么职业呢？"

In that last question, a surprising insight was revealed.

这个最后的问题，据透露出十分惊人的洞察力。

The vast majority of foreign fighters

绝大多数的外国战士

were seeking to become suicide bombers for martyrdom --

在谋求成为自爆烈士 — —

hugely important, since between 2003 and 2007, Iraq

极其重要的是，自 2003 年至 2007 年，伊拉克

had 1,382 suicide bombings, a major source of instability.

有 1,382 自杀性爆炸，不稳定的主要根源。

Analyzing this data was hard. The originals were sheets

分析这些数据是困难的。原始的表格

of paper in Arabic that had to be scanned and translated.

是阿拉伯语的，必须要扫描和翻译。

The friction in the process did not allow for meaningful

这过程中的摩擦不允许有意义的

results in an operational time frame using humans, PDFs

运用时间构架人类的结果，PDFs

and tenacity alone.

和独立的坚韧。

The researchers had to lever up their human minds

研究者们不得不撬动了他们人类的思维

with technology to dive deeper, to explore non-obvious

并运用科技去潜得更深，去探索非显而易见

hypotheses, and in fact, insights emerged.

的假说，事实上，目标实现了。

Twenty percent of the foreign fighters were from Libya,

20 ％ 的外国战士来自利比亚，

50 percent of those from a single town in Libya,

50%的人来自利比亚的同一个镇，

hugely important since prior statistics put that figure at

非常重要，因为以前的统计数字显示，

three percent. It also helped to hone in on a figure

是3%。它还帮助找出

of rising importance in al Qaeda, Abu Yahya al-Libi,

了基地组织新的重要人物，阿布叶海亚 · 阿尔-利比亚，

a senior cleric in the Libyan Islamic fighting group.

利比亚伊斯兰战斗组中的高级官员。

In March of 2007, he gave a speech, after which there was

2007 年 3 月，他给过一个演讲, 之后

a surge in participation amongst Libyan foreign fighters.

来自利比亚的外国战士数量激增。

Perhaps most clever of all, though, and least obvious,

也许最聪明的是，而且，最不明显的是，

by flipping the data on its head, the researchers were

通过整理其有关数据，研究人员

able to deeply explore the coordination networks in Syria

能够深入的探索了叙利亚的协调网络，

that were ultimately responsible for receiving and

它负责接收和

transporting the foreign fighters to the border.

运送外国战士到边境。

These were networks of mercenaries, not ideologues,

这些都是网络的雇佣军，不是空想，

who were in the coordination business for profit.

同过协调生意赚取利润。

For example, they charged Saudi foreign fighters

例如，他们收取沙特外国战士的钱

substantially more than Libyans, money that would have

大大超过利比亚的，钱

otherwise gone to al Qaeda.

最终会去向基地组织。

Perhaps the adversary would disrupt their own network

敌人可能会破坏他们的网络

if they knew they cheating would-be jihadists.

如果他们知道他们作弊是为了圣战。

In January, 2010, a devastating 7.0 earthquake struck Haiti,

2010 年 1 月，破坏性的 7.0级地震袭击了海地，

third deadliest earthquake of all time, left one million people,

有史以来第三次惨重的地震灾害，100 万人，

10 percent of the population, homeless.

10%的人口，无家可归。

One seemingly small aspect of the overall relief effort

一个看似小规模的整体救灾工作

became increasingly important as the delivery of food

变得越来越重要，从提供粮食

and water started rolling.

和水开始。

January and February are the dry months in Haiti,

在海地，一月和二月是干燥的月份

yet many of the camps had developed standing water.

然而许多难民的帐篷都被水淹了。

The only institution with detailed knowledge of Haiti's

唯一拥有海地详细信息的机构

floodplains had been leveled

已经被洪水淹没了

in the earthquake, leadership inside.

在地震中，包括领导。

So the question is, which camps are at risk,

所以，问题是，哪些营地处于危险之中，

how many people are in these camps, what's the

在这些营地有多少人，

timeline for flooding, and given very limited resources

洪水的时间是什么时候，给予有限的资源和时间

and infrastructure, how do we prioritize the relocation?

和基础设施，我们如何排定搬迁呢？

The data was incredibly disparate. The U.S. Army had

数据得出迥然不同的结论。美国陆军曾

detailed knowledge for only a small section of the country.

有一小部分这个国家详细资料。

There was data online from a 2006 environmental risk

在线数据源于2006年环境风险

conference, other geospatial data, none of it integrated.

会议，其他的地理空间数据，没有其它的集成。

The human goal here was to identify camps for relocation

在这里，人的目标是要确定可以搬迁的营地

based on priority need.

基于优先级别的需要。

The computer had to integrate a vast amount of geospacial

计算机不得不把大量的空间地理

information, social media data and relief organization

信息、 社交媒体数据和救援组织

information to answer this question.

信息集成后来回答这个问题。

By implementing a superior process, what was otherwise

通过实施一个优秀的程序，要不然

a task for 40 people over three months became

这是一个需要40 人超过三个月才能完成的任务

a simple job for three people in 40 hours,

现在简化到三个人用40小时就能完成的工作，

all victories for human-computer symbiosis.

这是人机共生所取得的胜利。

We're more than 50 years into Licklider's vision

Licklider 的梦想， 50 年之后

for the future, and the data suggests that we should be

的未来，数据表明我们应该

quite excited about tackling this century's hardest problems,

很兴奋的解决这个世纪最困难的问题，

man and machine in cooperation together.

人与机器在一起合作。

Thank you. (Applause)

谢谢。（掌声）

(Applause)

（掌声）