Kwabena Boahen: Making a computer that works like the brain

翻译人员: Mingzi Qu 校对人员: Xu Jiang

I got my first computer when I was a teenager growing up in Accra,

我成长在阿克拉（加纳首都），在我还是个少年的时候，我有了第一台电脑。

and it was a really cool device.

它真的是个很酷的玩意。

You could play games with it. You could program it in BASIC.

你可以用它来玩游戏，你可以用BASIC语言在上面编程。

And I was fascinated.

我被它迷住了。

So I went into the library to figure out how did this thing work.

于是我跑去图书馆想要弄明白这东西究竟是怎么工作的。

I read about how the CPU is constantly shuffling data back and forth

我了解到CPU（中央处理器）是如何不断地让数据来回穿梭于

between the memory, the RAM and the ALU,

存储器--RAM (随机存取存储器)和 ALU --

the arithmetic and logic unit.

算术逻辑运算器。

And I thought to myself, this CPU really has to work like crazy

我心想，CPU只有这样玩命地工作

just to keep all this data moving through the system.

才能让所有的数据在电脑系统中不停地运转呀。

But nobody was really worried about this.

但并没有人真正担心过这些。

When computers were first introduced,

当电脑首次问世时，

they were said to be a million times faster than neurons.

曾号称比人脑神经细胞快上一百万倍，

People were really excited. They thought they would soon outstrip

人们相当激动，他们以为电脑将很快就能超越

the capacity of the brain.

人脑。

This is a quote, actually, from Alan Turing:

Alan Turing 是这样说的：

"In 30 years, it will be as easy to ask a computer a question

“不出30年，向电脑提问就会变得和向人提问一样的

as to ask a person."

简单。”

This was in 1946. And now, in 2007, it's still not true.

这句话是在1946年说的。现在都2007年了，还是没能兑现。

And so, the question is, why aren't we really seeing

问题就在于，为什么我们不能真正地

this kind of power in computers that we see in the brain?

让电脑具备人脑的功能呢？

What people didn't realize, and I'm just beginning to realize right now,

过去大家都没意识到，而我也刚刚开始意识到的是

is that we pay a huge price for the speed

我们为了提升电脑的速度而付出了巨大的代价

that we claim is a big advantage of these computers.

这是因为速度被认为是电脑的一大优势。

Let's take a look at some numbers.

让我们看一些数字。

This is Blue Gene, the fastest computer in the world.

这是Blue Gene，世上最快的电脑。

It's got 120,000 processors; they can basically process

它拥有120，000个处理器；基本上它们每秒可以处理

10 quadrillion bits of information per second.

一万兆位元的信息。

That's 10 to the sixteenth. And they consume one and a half megawatts of power.

相当于10的16次方。并且它们还要消耗掉1.5兆瓦特的电力。

So that would be really great, if you could add that

如果你能把这些能量用到

to the production capacity in Tanzania.

坦桑尼亚的生产力上的话，那就简直棒极了。

It would really boost the economy.

它肯定能振兴经济。

Just to go back to the States,

再回来看看美国，

if you translate the amount of power or electricity

如果你把这个电脑消耗的电能换算成

this computer uses to the amount of households in the States,

美国家庭的用电量，

you get 1,200 households in the U.S.

那你会发现这相当于1200户美国家庭的用电量。

That's how much power this computer uses.

如此多的能量都被这个电脑消耗了。

Now, let's compare this with the brain.

现在，我们把这个电脑跟人脑做个比较，

This is a picture of, actually Rory Sayres' girlfriend's brain.

这是Rory Sayres 女友的大脑图片，

Rory is a graduate student at Stanford.

Rory 是斯坦佛大学的研究生，

He studies the brain using MRI, and he claims that

他用MRI（核磁共振成像）研究大脑，他宣称

this is the most beautiful brain that he has ever scanned.

这是他扫描过的最美丽的人脑。

(Laughter)

（笑声）

So that's true love, right there.

这就是真爱吧。

Now, how much computation does the brain do?

那么人脑究竟能计算多少呢？

I estimate 10 to the 16 bits per second,

我估计是每秒10到16位元

which is actually about very similar to what Blue Gene does.

这其实很接近Blue Gene (世界上最快的电脑）的运算能力了。

So that's the question. The question is, how much --

那么问题就在这儿。那就是——

they are doing a similar amount of processing, similar amount of data --

他们的计算量相似，处理的数据量相似——

the question is how much energy or electricity does the brain use?

可是人脑耗用了多少电能呢？

And it's actually as much as your laptop computer:

实际上就相当于你的笔记本电脑的用电量：

it's just 10 watts.

只有10瓦。

So what we are doing right now with computers

我们目前使用电脑做的事情

with the energy consumed by 1,200 houses,

消耗着相当于1200户家庭的总用电量，

the brain is doing with the energy consumed by your laptop.

而人脑做下来却只需要相当于笔记本电脑的用电量。

So the question is, how is the brain able to achieve this kind of efficiency?

那么问题是，大脑是怎么达到如此高效的？

And let me just summarize. So the bottom line:

让我总结一下，这个结论是：

the brain processes information using 100,000 times less energy

人脑用十万倍分之一的能量就处理了

than we do right now with this computer technology that we have.

我们目前的用电脑所处理的信息量。

How is the brain able to do this?

人脑是怎么做到这点的呢？

Let's just take a look about how the brain works,

我们先看看人脑是如何工作的，

and then I'll compare that with how computers work.

然后我同电脑是怎么工作的相比较。

So, this clip is from the PBS series, "The Secret Life of the Brain."

这个视频片段，剪辑自PBS（公共电视网）的系列片：“神秘的大脑”。

It shows you these cells that process information.

它向你展示了这些处理信息的细胞。

They are called neurons.

这些细胞叫做神经元。

They send little pulses of electricity down their processes to each other,

神经元之间在传送信息的过程中会发出微小的电脉冲，

and where they contact each other, those little pulses

神经元互相接触时，这些微小的电脉冲

of electricity can jump from one neuron to the other.

能够从一个神经元跳到另一个神经元。

That process is called a synapse.

这个过程被称为神经突触。

You've got this huge network of cells interacting with each other --

人脑所拥有的由神经元相互交织而成的网络相当庞大，

about 100 million of them,

其中有大概一亿个神经元，

sending about 10 quadrillion of these pulses around every second.

每秒发送约十万亿个这样的脉冲。

And that's basically what's going on in your brain right now as you're watching this.

你正观看这个片子的时候，你的大脑基本上就这样运转着。

How does that compare with the way computers work?

怎么拿这个跟电脑的工作方式比较呢？

In the computer, you have all the data

电脑把所有数据

going through the central processing unit,

都通过中央处理器来处理，

and any piece of data basically has to go through that bottleneck,

任何数据都必须通过那个瓶颈。

whereas in the brain, what you have is these neurons,

然而在大脑中，你所拥有的是这些神经元，

and the data just really flows through a network of connections

数据是顺着连接神经元的网络流动

among the neurons. There's no bottleneck here.

这里不存在瓶颈，

It's really a network in the literal sense of the word.

这是一个名副其实的网络，

The net is doing the work in the brain.

就是这个网络担负着大脑的运转。

If you just look at these two pictures,

看看这两张图片，

these kind of words pop into your mind.

你的脑海中会跳出这样的词，

This is serial and it's rigid -- it's like cars on a freeway,

这一幅连续又呆板：就像在高速路上的汽车——

everything has to happen in lockstep --

一切必须按部就班；

whereas this is parallel and it's fluid.

而这幅图平行而且有流动感，

Information processing is very dynamic and adaptive.

其信息处理既非常活跃又很具适应性。

So I'm not the first to figure this out. This is a quote from Brian Eno:

我并不是第一个有这样想法的人。Brian Eno如是说：

"the problem with computers is that there is not enough Africa in them."

“电脑的问题就在于它还不能大到足以装下整个非洲。”

(Laughter)

（笑声）

Brian actually said this in 1995.

Brian 说这话时是1995年

And nobody was listening then,

当时没有任何人听进去，

but now people are beginning to listen

但现在人们开始当真了

because there's a pressing, technological problem that we face.

因为我们正面临着一个急迫的技术问题

And I'll just take you through that a little bit in the next few slides.

在下面几张幻灯片中我会带你们简略地了解一下这个问题：

This is -- it's actually really this remarkable convergence

这个——实际上它是个真正非凡的聚合体

between the devices that we use to compute in computers,

算是介于用电脑来计算的装置

and the devices that our brains use to compute.

和用人脑来计算的装置之间。

The devices that computers use are what's called a transistor.

电脑用的装置我们称之为晶体管。

This electrode here, called the gate, controls the flow of current

这儿是电极，称为闸极，控制着从源极流向汲极的电流

from the source to the drain -- these two electrodes.

就是这两个电极。

And that current, electrical current,

而那电流，电流是由电子传送的，

is carried by electrons, just like in your house and so on.

就如同你房子里的电流一样，诸如此类。

And what you have here is, when you actually turn on the gate,

这里你要明白的是，当你打开闸极时，

you get an increase in the amount of current, and you get a steady flow of current.

电流量会增加，并且是股稳定电流。

And when you turn off the gate, there's no current flowing through the device.

当你关掉闸极时，就没有任何电流流过这个装置了。

Your computer uses this presence of current to represent a one,

我们用“一”来代表你的电脑有电流，

and the absence of current to represent a zero.

而用“零”来代表电脑无电流。

Now, what's happening is that as transistors are getting smaller and smaller and smaller,

现在的情况是当晶体管变得越来越小越来越小的话，

they no longer behave like this.

它们就不会再像这样运转了，

In fact, they are starting to behave like the device that neurons use to compute,

事实上，它们会开始像神经元用来计算的装置那样来运作，

which is called an ion channel.

这种装置被称作离子通道

And this is a little protein molecule.

这是个小小的蛋白质分子。

I mean, neurons have thousands of these.

我的意思是，神经元有成千上万个这样的分子。

And it sits in the membrane of the cell and it's got a pore in it.

而它就分布在细胞膜中并且自身还带了个小孔，

And these are individual potassium ions

而这些是单独的钾离子，

that are flowing through that pore.

在小孔中穿来过去。

Now, this pore can open and close.

现在，这个孔是能开能闭的，

But, when it's open, because these ions have to line up

但是，当它打开时，由于这些离子必须排成一行

and flow through, one at a time, you get a kind of sporadic, not steady --

一个一个地穿过小孔，因此产生零星的，而非稳定的——

it's a sporadic flow of current.

一股断断续续的电流。

And even when you close the pore -- which neurons can do,

甚至当小孔闭合的时候——神经元做得到这点的，

they can open and close these pores to generate electrical activity --

它们可以通过开关这些小孔来产生电活动——

even when it's closed, because these ions are so small,

甚至当孔闭合时，由于这些离子如此之小，

they can actually sneak through, a few can sneak through at a time.

它们实际上可以偷偷地穿过，其中一些还可以同时偷偷地穿过。

So, what you have is that when the pore is open,

所以你得出的结论就是当小孔张开时，

you get some current sometimes.

有时候会产生一些电流

These are your ones, but you've got a few zeros thrown in.

这些就是“一”，但你也会额外得到些“零”；

And when it's closed, you have a zero,

而当小孔闭合时，你则得到个“零”，

but you have a few ones thrown in.

但你也能得到些额外的“一”，是的。

Now, this is starting to happen in transistors.

现在，这一原理正开始运用于晶体管，

And the reason why that's happening is that, right now, in 2007 --

而发生的原因就在于，在2007年，

the technology that we are using -- a transistor is big enough

我们目前使用的技术，晶体管的大小足以使

that several electrons can flow through the channel simultaneously, side by side.

好几个电子能同时穿过通道，而且是并排地。

In fact, there's about 12 electrons can all be flowing this way.

事实上，大概有12个电子都可以这样穿过去。

And that means that a transistor corresponds

这意味着一个晶体管就相当于

to about 12 ion channels in parallel.

12个并列的离子通道。

Now, in a few years time, by 2015, we will shrink transistors so much.

在几年时间内，到2015，我们会把晶体管体积缩得非常小，

This is what Intel does to keep adding more cores onto the chip.

这正是英特尔所致力于的事业：不断地往芯片上

Or your memory sticks that you have now can carry one gigabyte

或是你的记忆棒上添加更多的核，这样就使它们能有1G的内存

of stuff on them -- before, it was 256.

——而以前才256MB。

Transistors are getting smaller to allow this to happen,

晶体管变得越来越小才使这一切得以实现

and technology has really benefitted from that.

而技术更是得益于此

But what's happening now is that in 2015, the transistor is going to become so small,

但现在的情况是到2015年，晶体管将变得如此之小，

that it corresponds to only one electron at a time

以至于它一次只能让一个电子

can flow through that channel,

通过通道，

and that corresponds to a single ion channel.

相当于一个单离子通道，

And you start having the same kind of traffic jams that you have in the ion channel.

因此就会开始出现像在单离子通道里发生的那种交通堵塞，

The current will turn on and off at random,

电流会随机地时断时续，

even when it's supposed to be on.

甚至在它理应通电的时候

And that means your computer is going to get

而那也就意味着你的电脑将会把

its ones and zeros mixed up, and that's going to crash your machine.

它的“一”和“零”们搞混淆，那么你这台机器就完蛋了。

So, we are at the stage where we

所以，我们现在还处于

don't really know how to compute with these kinds of devices.

尚不真正清楚如何使用这类装置来运算的阶段

And the only kind of thing -- the only thing we know right now

而我们目前知道的唯一一件事，唯一

that can compute with these kinds of devices are the brain.

能用这类装置来进行运算的是我们人类的大脑

OK, so a computer picks a specific item of data from memory,

好吧，所以说电脑是从内存中中挑取特定的一项数据

it sends it into the processor or the ALU,

把它传送给处理器或是ALU，

and then it puts the result back into memory.

然后再将运算结果送还给内存。

That's the red path that's highlighted.

就是这着重标出的红色路线。

The way brains work, I told you all, you have got all these neurons.

人脑工作的方式，用上了你能有的所有神经元。

And the way they represent information is

它们呈现信息的方式是

they break up that data into little pieces

把数据粉碎成很小的碎片

that are represented by pulses and different neurons.

并用脉冲和不同的神经元来表达。

So you have all these pieces of data

而所有的数据碎片

distributed throughout the network.

都分散在这网络中。

And then the way that you process that data to get a result

而你处理数据得到结果的方式则是

is that you translate this pattern of activity into a new pattern of activity,

将这一种活动模式转化成一种新的活动模式，

just by it flowing through the network.

仅仅就是让它在网络中流过而已。

So you set up these connections

这样你就建立起这些连接，

such that the input pattern just flows

仅仅让输入模式流动

and generates the output pattern.

就能产生输出模式

What you see here is that there's these redundant connections.

现在你看到的是一堆多余的连接

So if this piece of data or this piece of the data gets clobbered,

所以如果这块信息碎片，或这块信息碎片被损毁的话

it doesn't show up over here, these two pieces can activate the missing part

它就不会在这儿显示出来了，这两份信息可以

with these redundant connections.

通过这些多余的连接来激活缺失的部分信息

So even when you go to these crappy devices

所以即使你用的是这么些蹩脚的装置

where sometimes you want a one and you get a zero, and it doesn't show up,

有时会在你想要个一的时候给你个零

there's redundancy in the network

网络中的重复性

that can actually recover the missing information.

实际上却能恢复那些缺失的信息。

It makes the brain inherently robust.

它令大脑自然而然地强大。

What you have here is a system where you store data locally.

你这里所拥有的是一个只能存储本地数据的系统

And it's brittle, because each of these steps has to be flawless,

而且很脆弱，因为它的每一个步骤都必须是准确无误的

otherwise you lose that data, whereas in the brain, you have a system

否则你就会丢失数据。然而大脑系统

that stores data in a distributed way, and it's robust.

以分散的方式存储数据，且强大无比。

What I want to basically talk about is my dream,

我想讨论的基本问题是我的梦想，

which is to build a computer that works like the brain.

那就是建造一个像大脑那样工作的电脑。

This is something that we've been working on for the last couple of years.

过去数年来我们一直在为此而努力。

And I'm going to show you a system that we designed

而现在我将向你们展示一个我们设计的

to model the retina,

模拟视网膜的系统

which is a piece of brain that lines the inside of your eyeball.

这一模拟系统就是覆盖在你眼球内部的一层大脑

We didn't do this by actually writing code, like you do in a computer.

实际上我们做的这一模拟系统并不像做电脑系统那样编程，

In fact, the processing that happens

事实上，这一小片大脑的

in that little piece of brain is very similar

运作过程非常类似于

to the kind of processing that computers

电脑从因特网上获取视频流的

do when they stream video over the Internet.

过程。

They want to compress the information --

人们想要压缩信息——

they just want to send the changes, what's new in the image, and so on --

人们只想发送新发生改变的图像，诸如此类——

and that is how your eyeball

而那就是你的眼球如何

is able to squeeze all that information down to your optic nerve,

能把所有捕捉到的信息传送到你的视神经

to send to the rest of the brain.

再传送给大脑的其余部分

Instead of doing this in software, or doing those kinds of algorithms,

取代了用软件或者做各种各样的算法来做这一系统

we went and talked to neurobiologists

我们去问了神经生物学家

who have actually reverse engineered that piece of brain that's called the retina.

他们用反工程法解析了被称作视网膜的那片大脑。

And they figured out all the different cells,

而且他们分析出所有不同的细胞

and they figured out the network, and we just took that network

还分析出其网络，我们只是拿着那个网络

and we used it as the blueprint for the design of a silicon chip.

用它作为设计硅片的蓝图。

So now the neurons are represented by little nodes or circuits on the chip,

现在硅片上的小结点或电路代表神经元，

and the connections among the neurons are represented, actually modeled by transistors.

神经元之间的连接实际上由晶体管模拟

And these transistors are behaving essentially

这些晶体管的运作模式基本上

just like ion channels behave in the brain.

就如同大脑中的离子通道的运作模式。

It will give you the same kind of robust architecture that I described.

这是同我描述过的一样的强大结构。

Here is actually what our artificial eye looks like.

这就是我们的人造眼睛的样子。

The retina chip that we designed sits behind this lens here.

我们设计的视网膜硅片安置在这里的镜片后。

And the chip -- I'm going to show you a video

而这硅片——我将给你们看一段视频

that the silicon retina put out of its output

是这个硅片视网膜的输出

when it was looking at Kareem Zaghloul,

当它看着Kareem Zaghloul 的时候，

who's the student who designed this chip.

Kareem是设计这块硅片的学生。

Let me explain what you're going to see, OK,

让我解释一下你将看见什么，好吗？

because it's putting out different kinds of information,

由于输出各种不同信息

it's not as straightforward as a camera.

它不像相机那么简单。

The retina chip extracts four different kinds of information.

视网膜硅片摄取四种不同的信息。

It extracts regions with dark contrast,

它摄取黑色对比区域，

which will show up on the video as red.

在视频上表现为红色。

And it extracts regions with white or light contrast,

它也摄取白色或亮色对比区域，

which will show up on the video as green.

在视频上显示为绿色。

This is Kareem's dark eyes

这是Kareem的黑眼睛

and that's the white background that you see here.

而这儿是你所看见的白色背景。

And then it also extracts movement.

然后硅片也摄取物体的运动。

When Kareem moves his head to the right,

当Kareem把头转向右边，

you will see this blue activity there;

你能看见那儿的蓝色活动区域，

it represents regions where the contrast is increasing in the image,

它代表图像中对比加强的区域，

that's where it's going from dark to light.

这一区域由暗变明。

And you also see this yellow activity,

而且你也看见这块黄色活动区域，

which represents regions where contrast is decreasing;

它代表对比逐渐减弱区域，

it's going from light to dark.

这一区域由亮变暗。

And these four types of information --

而这四种信息类型——

your optic nerve has about a million fibers in it,

你的视神经约有一百万根神经纤维，

and 900,000 of those fibers

这些神经纤维中的90万根

send these four types of information.

传送这四种信息。

So we are really duplicating the kind of signals that you have on the optic nerve.

所以我们真正地复制了视神经上的那类信号。

What you notice here is that these snapshots

你这里所注意到的是这些

taken from the output of the retina chip are very sparse, right?

从视网膜硅片的输出信息所摄取的快照是非常分散的。

It doesn't light up green everywhere in the background,

在背景中并非到处都是亮色调的绿色，

only on the edges, and then in the hair, and so on.

仅仅在边缘如此，等等。

And this is the same thing you see

而这同你所看见的一样

when people compress video to send: they want to make it very sparse,

当人们压缩视频后发送：他们想把它做得很分散，

because that file is smaller. And this is what the retina is doing,

因为文件更小了。而这就是视网膜所做的一切

and it's doing it just with the circuitry, and how this network of neurons

仅仅用电路就做到了，而且这个神经网络是如何

that are interacting in there, which we've captured on the chip.

在那儿相互作用的，我们都在硅片上捕捉到了。

But the point that I want to make -- I'll show you up here.

但是我要说的是，看这里。

So this image here is going to look like these ones,

这样的图像将看上去像那些图像,

but here I'll show you that we can reconstruct the image,

但在这儿我将向你们演示我们能重组图像,

so, you know, you can almost recognize Kareem in that top part there.

所以，你知道，你们几乎可以在那幅顶部图像分辨出Kareem.

And so, here you go.

瞧，就是这个。

Yes, so that's the idea.

是的，这就是我的想法。

When you stand still, you just see the light and dark contrasts.

当你站着不动时，你只看见明暗对比。

But when it's moving back and forth,

但是当你前后移动时，

the retina picks up these changes.

视网膜就摄取到这些变化。

And that's why, you know, when you're sitting here

那就是为什么，当你坐在这儿，

and something happens in your background,

在你的背后发生变化时，你也能知道，

you merely move your eyes to it.

你只要需要看一眼。

There are these cells that detect change

这些细胞探测到变化

and you move your attention to it.

你就把注意力转向它。

So those are very important for catching somebody

这对你发现

who's trying to sneak up on you.

想对你偷偷摸摸的家伙非常重要。

Let me just end by saying that this is what happens

让我这么说吧，作为这次演讲的结束，这就是

when you put Africa in a piano, OK.

当你把非洲放入一架钢琴会发生的一切，好吧。

This is a steel drum here that has been modified,

这是一架已被改装的钢鼓，

and that's what happens when you put Africa in a piano.

而那是你把非洲放入钢琴所发生的事情。

And what I would like us to do is put Africa in the computer,

而我想让大家所做的是把非洲放入一台电脑，

and come up with a new kind of computer

变出一种新电脑

that will generate thought, imagination, be creative and things like that.

这种电脑将产生思想，具有想象力，充满创造力，以及诸如此类的能力。

Thank you.

谢谢诸位。

(Applause)

（掌声）

Chris Anderson: Question for you, Kwabena.

Chris Anderson:Kwabena, 有一个问题问你，

Do you put together in your mind the work you're doing,

你是否想过，在你正从事的工作

the future of Africa, this conference --

非洲的未来，和这次大会——

what connections can we make, if any, between them?

在它们之间我们可以获得什么联系，如果有的话？

Kwabena Boahen: Yes, like I said at the beginning,

Kwabena Boahen:是的，正如我一开始所说的，

I got my first computer when I was a teenager, growing up in Accra.

我在阿克拉长大，在青少年时期有了第一台电脑。

And I had this gut reaction that this was the wrong way to do it.

我的本能反应是这种做法是错误的。

It was very brute force; it was very inelegant.

这是一种非常不理性的力量，非常不雅。

I don't think that I would've had that reaction,

我不认为我会有如此的反应，

if I'd grown up reading all this science fiction,

如果我从小读着所有这些科幻小说，

hearing about RD2D2, whatever it was called, and just -- you know,

听着有关星球大战中的机器人RD2D2，不管你怎么称呼它，只是--你知道的，

buying into this hype about computers.

认同对电脑的这种炒作。

I was coming at it from a different perspective,

我是从一个不同的视角接触电脑的，

where I was bringing that different perspective

我正是带这这个不同的视角

to bear on the problem.

来承受这个问题。

And I think a lot of people in Africa have this different perspective,

而且我认为非洲许许多多人有这种不同的观点，

and I think that's going to impact technology.

而且我认为这将影响到技术。

And that's going to impact how it's going to evolve.

影响到技术的发展的方向。

And I think you're going to be able to see, use that infusion,

而且我认为你们将能目睹，利用那种新灌输的思想，

to come up with new things,

来创造新事物，

because you're coming from a different perspective.

因为你们来自于一个不同的背景。

I think we can contribute. We can dream like everybody else.

我觉得我们可以像其他任何人一样贡献自己的力量，构筑自己的梦想。

CA: Thanks Kwabena, that was really interesting.

Chris Anderson:谢谢，Kwabena,这真的很有意思。

Thank you.

谢谢。

(Applause)

（掌声）