与摩根·弗里曼一起穿越虫洞（2010）

Through The Wormhole With Morgan Freeman S04E07

Part 1 April 19, 2021, 2:53 p.m.

我们正处于一个潜在的变革之中
We are in the midst of a revolution so insidious,
而我们却看不到变化
we can't even see it.
机器人存在于我们身边并为我们工作
Robots live and work beside us.
现在我们要让它们学会思考自我
And now we're designing them to think for themselves.
我们正在赋予它们凭借自己学习进步的力量
We're giving them the power to learn to move on their own.
这些新的生命形式会变得
Will these new life-forms evolve
比我们更聪明更能干吗？
to be smarter and more capable than us?
还是我们会和机械融为一体？
Or will we choose to merge with the machines?
机器人是人类进化的未来吗？
Are robots the future of human evolution?
空间，时间，生命本身
Space, time, life itself.
宇宙之谜，尽在穿越虫洞
The secrets of the cosmos lie through the wormhole.
与摩根·弗里曼一起穿越虫洞（S04E07）
Through The Wormhole With Morgan Freeman (S04E07)
机器人是人类进化的未来吗？
Are Robots the Future of Human Evolution?
我们人类喜欢认为自己
We humans like to think of ourselves
处于进化的顶端
as the pinnacle of evolution.
我们确实是地球上最聪明的、最具适应力的生命形式
We are the smartest, most adaptable form of life on earth.
我们已经重塑这个世界来满足我们的需要
We have reshaped the world to suit our needs.
就像智人替代了直立猿人
But just as homo sapiens replaced homo erectus,
必然会有什么将会替代我们
it's inevitable something will replace us.
但假若我们能够自己制♥造♥继承人呢？
What if we're building our own successors?
就像我们学习走动、思考、感受自我一样
Just as we learn to move, think, and feel for ourselves,
我们现在正在给予机器人同样的能力
we're now giving robots those same powers.
这将会把我们带到什么样的境地？
Where will this lead?
这是人类将会转变的形式吗？
Is this what humanity will become?
少年时，我用一些备用部件造了一辆自行车
When I was a teenager, I built a bicycle from spare parts.
我造的自行车平衡性非常好
My bicycle was so well-balanced,
我可以推动它就保持平衡而不需要骑在上面
I could jog alongside it without holding onto it.
没人认为这怎么厉害，但却使我思考
Nobody was that impressed, but it made me think.
我们有一天会拥有可以
Would we one day have machines
完全凭借自身行动的机器吗？
that truly could move on their own?
那个时候它们还会需要我们吗？
Would they even need us anymore?
剑桥大学的丹尼尔·沃尔珀特
Daniel Wolpert of the university of Cambridge
认为如果机器人
believes that if robots
是人类进化的未来
are to be the future of human evolution,
它们就必须像我们一样学习行动
they're going to have to learn to move as well as we do...
因为行动是我们强大智慧的
Because movement is the supreme achievement
最高成就
of our powerful intellect.
我认为我们能够提出的最基本的问题是
The most fundamental question I think we can ever ask is,
为什么以及何时动物进化出了脑？
why and when have animals ever evolved a brain?
现在，当我问我的学生这个问题时
Now, when I ask my students this question,
他们会告诉我
they'll tell me,
我们用大脑来思考或感知世界
we have ones to think or to perceive the world,
但这是完全错误的
and that's completely wrong.
我们只因为一个原因而具有大脑
We have a brain for one reason and one reason only,
那就是要产生能够适应环境的复杂行为
and that's to produce adaptable and complex movement,
因为行为才是我们具有的
because movement is the only way we have
唯一可以影响世界的方式
of affecting the world around us.
我们大脑的各种智力
All of our brains' intellectual capacity
都源于一个原始的动机
grew from one primal motivation --
学得如何更加擅长行动
to learn how to move better.
这包括我们用两条腿行走的能力
It was our ability to walk on two legs,
使用复杂的面部表情交谈和表露情感
to speak and emote with complex facial movements,
并灵巧的控制四肢
and to manipulate our dexterous limbs
使人类站在食物链顶端
that put humans on top of the food chain.
如果没有行动的能力
There can be no value
感知、情感和思考
to perception or emotions or thinking
都没有意义
without the ability to act.
所有的其他特征
All those other features,
比如记忆、认知、爱、恐惧都要发展为行为
like memory, cognition, love, fear play into movement,
这些都是大脑输出的最终结果
which is the final output of the brain.
没有任何机器可以做出
No machine could handle
像我们每天做的多种复杂行为
the huge variety of complex movements we perform every day.
想象一个机器人
Just imagine a robot
想要玩一场英格兰最流行的消遣游戏
trying to play one of England's most famous pastimes.
尽管这次击球看起来很简单
So, although that shot looks simple
对于我来说毫不费力
and it felt effortless to me,
而我脑中进行的则是非常复杂的
the complexity of what's going on in my brain
肢体协调
is really quite remarkable.
我必须注意投球手投出的球
I have to follow the ball as the bowler bowls it
并预测它会弹到哪里
and predict where it's going to bounce
以及它如何从地面弹起
and how it's gonna rise from the ground.
然后我需要作出决定
I then have to make a decision
我要如何去击球
as to what type of shot I'm going to make.
最终，我需要协调600块肌肉
And finally, I have to contract my 600 muscles
通过特定的顺序来进行击球动作
in a particular sequence to execute the shot.
此时，每一块肌肉的运动
Now, each of those components
都经过了复杂的数学♥运♥算
has real mathematical complexity,
而这种能力现今远超所有的机器人
which is currently beyond the ability of any robotic device.
让机器人拥有
One of the greatest challenges
我们这样行为的最大挑战之一
in getting robots to move like we do
就是要让它们学会处理不确定因素
is teaching them to deal with uncertainty --
而对于我们的大脑来说这是本能
something our brains do intrinsically.
球不会以两次相同的情况飞向你
A ball will never come at you the same way twice.
每次你都必须要迅速调整动作
You must instantly adjust your swing each time.
问题是
The question is...
人脑如何处理这些不确定因素？
How does the human brain deal with all this uncertainty?
丹尼尔认为这需要一种概率估计理论
Daniel thinks it uses a theory of probability estimation
称作贝叶斯推理
called bayesian inference
才能进行运算
to figure it out.
击球手现在要做的很重要的事情
So, a critical thing the batsman now has to do
就是这个球会在哪里弹起
is decide where this ball is going to bounce,
只要他们能够准确击球
so as they can prepare the correct shot,
就需要贝叶斯推理
and for that, they need bayesian inference.
贝叶斯推理可以做的是
What bayesian inference is all about
从两种不同的信息来源中
is deciding, optimally, the bounce location of the ball
决定球会弹起的最准确的位置
from two different sources of information.
一个信息来源是很明显的
One source of information is obvious.
就是你盯着球
You look at the ball.
你可以通过观察球飞来的
So, you can use vision of the trajectory of the ball
运动轨迹
as it comes in
来尝试估计它会在哪里弹起
to try and estimate where it's going to bounce.
但是视觉并不完美
But vision is not perfect,
因为我们对视觉的加工存在变动
in that we have variability in our visual processors,
所以至少落地点的变动范围就是红色圆圈显示的
so at least where distribution's shown in red here
作为球会落在某一点的概率分布
as the probable bounce locations.
但贝叶斯规则说还有一个信息来源
But bayes' rule says there's another source of information.
就是关于落地点的先验知识
It's the prior knowledge about possible bounce locations.
如果你是个好击球手
If you're a good batter,
你就能有效地看着投球手
then you can effectively look at the bowler
就能通过他特殊的投球方式或暗示了解到球的运行方式
and maybe know by his particular bowling style or small cues --
这里用蓝色底纹表示
and that's shown by the blue shading --
这是一个不同的区域
which is a different area.
那么贝叶斯推理就整合了
So, bayesian inference is a way of combining
红色范围和蓝色范围
this red distribution with the blue distribution,
你通过两种信息的交集
and you do that by multiplying the numbers together in each
产生了这个黄色分布
to generate this yellow distribution,
这个分布称为可信度
which is termed the belief.
通过使用这样的信息
And using that information,
击球手现在就可以准备击球了
the batsman can now prepare his shot.
好吧，我应该躲开一点
Okay, I should probably get out of the way.
击球手的大脑，就像我们所有人
The batsman's brain, like all of ours,
自动进行了这些计算
is doing this math automatically.
我们这个物种锤炼出了对于行为的预测
We are a species that's honed for movement prediction.
这就使我们成为了
It's what has made us
这颗星球上最好的猎人和工具匠
the planet's best hunters and toolmakers.
我们已经有了比我们行动
We already have robots that are faster and more accurate
更快更准确的机器人
than we are.
但是我们要为它们的每个行为编程
But we have to program their every move.
机器人只能沿着我们已经走过的
For robots to walk down the evolutionary road
进化道路前进
we've already traveled,