人工智能的真相（2023）（3）

我已有三十年驾龄
I've been a driver for 30 years,
但从未见过像这样的车
but I've never been in a car like this one before.
这是辆自动驾驶汽车
This is an autonomous vehicle
正自己行驶在牛津的街道上
and it's driving itself through Oxford.
已开启自动驾驶
在千禧年前后 现实生活可用的无人驾驶汽车
At the turn of the century, practical driverless cars
似乎还是一场非常非常遥远的梦
felt like they were a very, very distant dream.
但人工智能和计算机技术的进步
But advances in AI and computing
让无人驾驶汽车指日可待
mean that they could be just around the corner.
我们之所以憧憬无人驾驶汽车 是因为它们
And the dream of driverless cars is that they will be far safer
比人类司机更安全 更高效
and more efficient than human drivers could ever be.
但有人担心
But some people worry
我们可能把太多控制权
that we might end up giving too much control
交给人工智能
to artificial intelligence.
会不会有什么隐患或不可预见的后果
Could there be any hidden dangers, any unforeseen consequences?
人工智能会威胁到我们的存在吗
And could AI even threaten our existence?
假象
真♥相♥
人工智能大揭秘
2023年英国皇家科学院圣诞讲座
第三讲 人工智能的未来: 美梦还是噩梦
欢迎来到第三场也是最后一场
Welcome to the third and final
麦克·伍德里奇教授
英国皇家科学院圣诞讲座
Christmas lecture from the Royal Institution,
本节目由CGI集团赞助播出
an event supported by CGI.
本次讲座 我们将探讨人工智能的前景
In this lecture, we'll be looking at the promise of AI
及其在未来生活中的作用
and its role in your future.
我们是该感到兴奋还是担忧
Should we be excited or should we be concerned?
或者两者兼有之
Or maybe a little bit of both?
都有可能出现什么问题
How might things go wrong
我们要怎样做 才能确保不出错
and what can we do to make sure that that doesn't happen?
随着人工智能变得越来越聪明
And as AI becomes ever smarter,
它最终会产生意识吗
could it ever actually be conscious
机器的意识和我们的一样吗
and would machine consciousness be the same as ours?
还是说 意识为人类独有
Or are human beings unique?
我们先听听DeepMind的联合创始人
Let's start by seeing what Demis Hassabis,
德米斯·哈萨比斯 对人工智能未来的看法
the co-founder of DeepMind, thinks about the future of AI.
我认为能在这个时代从事科技工作
I think it's an incredibly exciting time
这是非常激动人心的
to be in technology and science.
我将人工智能视作
I see AI as like the ultimate tool
帮助各位专家和科学家的终极工具
to assist experts and scientists with their endeavours.
我想给初入职场的人一些忠告 那就是
And what I would say for anyone starting out now in their career
你们真的应该深入研究这些系统的工作原理
is to really deep dive into how these systems work
学习如何利用这些系统
and how they can utilise these systems
来帮助自己工作
to help them in their own work.
我认为人工智能工具
I think it's almost limitless,
在工作中的应用有着无限的潜力
the potential for utilising these tools and in your own workflows.
但它每时每刻
But it's going to be, you know,
都在更新换代
it's going to be changing every moment.
所以我想说的另一件事是 学习
So I think one other thing I would say is to learn, like,
迅速学习并适应接下来发生的一切
learn really fast and be adaptive to whatever's coming next.
因为唯一能确定的是
Cos the only thing that's sure is,
未来将有许多变化
there's going to be a lot of change.
让我们回到之前展示的
So let's go back to that driverless
无人驾驶汽车之旅
car journey that we showed you earlier,
我们将看到人工智能
and we're going to see how AI
如何应对我家乡牛津狭窄拥挤的街道
dealt with the mean streets of my hometown, Oxford.
下面有请牛津大学的保罗·纽曼教授
Please welcome, from Oxford, Professor Paul Newman.
欢迎你 保罗 -谢谢
Welcome, Paul. - Thank you.
你已经乘坐过无数次无人驾驶汽车了
Now, you've been in driverless cars a million times.
而我是第一次乘坐
That was my first time
真是非常奇怪的经历
and it was a really weird experience.
我们在牛津遇到的路况相当复杂
We went through some quite complicated traffic in Oxford
汽车的驾驶席却空无一人
and there was nobody behind the wheel.
那么 汽车是如何知道车身周围的情况呢
So how did the car know what was going on around it?
这辆车装备有各种传感器
Well, the vehicle's equipped with various sensors.
有摄像头 还有激光探测器
It has cameras, it has lasers.
从这个片段可以看到
And you can see those cameras there
人工智能正在处理摄像头的画面
being having their images processed by an AI,
这是真实旅程吗 -追踪人们的...
This is our actual journey? - tracking people...
是的 这是牛津大街上的真实旅程
This is your actual journey, yes, up through Oxford.
你可以看到它在通过识别图像 追踪周围的汽车
And you can see it's tracking cars, identifying from images,
在路口还能识别红绿灯
traffic lights, where there are junctions,
所以它知道周围有什么及如何驾驶
so it knows how to move and what's around it.
我们有摄像头能看到
So we have cameras that can see
并捕捉图像中的不同颜色
and pick up different colours in images.
我们有激光探测器能测量与物体的距离
We have lasers that can measure the distance to things,
有雷达能看到摄像头和激光探测器
and we have radars that can see perhaps where lasers and cameras
在浓雾中无法看清的东西
may have trouble seeing through fog.
有这许多不同的传感器 就能确保它知道
So lots of different sensors here have a robust way of knowing
你的具体♥位♥置 -好吧
where you are in the world. - Ok.
所有这些传感器提供了
So all of those sensors, they're providing
一连串你周围环境的数据
a barrage of dataabout your environment
你们都会用这些信息做什么呢
and what are you actually doing with that information?
这么做用意何在
What's the point of it?
无人驾驶汽车
There are three key questions
需要解决三个关键问题
that the vehicles need to answer in the vehicle.
第一 我在哪里
One, where am I?
第二 我周围是什么
Two, what's around me?
我周围有什么东西
What stuff is around me?
第三 那些东西会做出什么行为
Three, what's going to happen with that stuff
我又该如何应对
and what should I do because of that?
对即将发生事情做出预测后 我又该怎样反应
How should I drive because of my predictions of what's happening?
最近的问题是 这些车应该共享哪些数据
And more recently, then, is what data should those vehicles share
以便训练其他无人驾驶汽车
so they can train other vehicles?
好吧 以我的经验
OK, now, in my experience,
开车最难的部分莫过于
the tough thing about driving is other human beings.
道路上的其他人 无论他们是驾车
It's other road users, whether they're in cars,
骑自行车 还是步行
on bicycles or just pedestrians.
当你的车遇到突发事件时
So how does your car decide what to do
是如何决定怎么做的
when it encounters something unpredictable?
我们有段之前行驶时
I think we have a video of something
发生类似情景的录像
that happened while we were driving.
好
OK, yes.
你们来到一个环岛 一辆出租车超车
So you come up to a roundabout and a car passes
与你们擦身而过
and that taxi was very close.
当时车载人工智能所做的
So what the vehicle was doing at that point,
就是对可能发生的事做出许多假设
the AI, is we're making lots of hypotheses of what could happen.
那辆出租是会停车 还是靠得太近
The vehicle was going to pull in, get too close.
还是会让你先走
It was going to let you go in front.
许多假设
Lots of things.
好吧 你们从视频中看不出来
OK, what you don't get from that video
当时我们正在过一个环岛
is we were going round a roundabout,
那辆车就从夹缝间穿过
this car really drove in the middle.
实际上 假如当时是我在开车
And, actually, if I had been a human driver behind a wheel,
那将是非常恐怖的经历
that would have been a very scary experience.
但这辆车只是缓缓地踩下了刹车
But the car just seamlessly put its brake on and you hardly,
你几乎都注意不到 真是太不可思议了
you hardly noticed, it was really incredible.
保罗 我们听说在其他国家
Now, we hear in other countries, Paul,
有更多无人驾驶汽车
there are many more driverless cars,
比如在美国等国家的街头
for example, on the streets of the United States and so on.
我们也听说发生了些事故
And we hear about accidents,
还有人员受伤 甚至死亡
we hear about injuries and fatalities and so on.
自动驾驶汽车在街道上行驶真的安全吗
Is it really safer to have autonomous vehicles on our streets?
我想会很安全
I think it will be.
我认为有充分的理由相信
And I think there's a really very strong reason
这项技术
why this technology
将对交通运输产生巨大影响
is going to make an extraordinary difference to transport.
作为人类 当你学开车时
So as you as a human, when you learn to drive,
你只有自己的经验
you only have your own experience.
但这些自动驾驶汽车就不同了
But that's going to be so different for autonomous vehicles,
它们可以向所有其他有行驶经验的车学习
that will learn to drive from every other vehicle that has driven,
基于人工智能 我们能让这些自动驾驶汽车
cos these AIs, we can have them share every mile
分享行驶的每一英里