评论
打印
There are lots of different types of robots; insect robots, animal robots, jellyfish robots, toy robots, cooking robots, military robots, sex robots?
机器人的种类多种多样：什么昆虫机器人、动物机器人、水母机器人、玩具机器人、烹饪机器人、军事机器人，乃至性机器人？
An android is a human-like robot. It looks like a human, talks like a human, it acts like a human.
安卓是一种类人机器人。它长得像人，说话像人，行动起来也像人。
Androids are the pinnacle of robotics and artificial intelligence.
安卓是机器人学和人工智能的巅峰之作。
Successfully building an android is essentially like creating synthetic life in our own image.
成功建造安卓，基本就像是依照我们自身的形象，创造合成生命。
So how close are we to building an android?
那么我们距离建造安卓有多远呢？
Right now there are only a handful of robots that are anywhere close to passing themselves off as human.
目前类人机器人仍屈指可数。
Not only do these 'bots look and feel weird, their artificial intelligence systems aren't that intelligent.
这些机器人不仅看起来行为怪异，而且他们的人工智能系统也不那么智能。
But robotics and computer engineers are working on humanoid, intelligent robots all over the world. Why?
但是研究类人机器人，智能机器人的机器人技术和计算机工程师，遍布世界各地。为什么呢？
Because who doesn't want a robot cook, butler, or driver?
因为谁不想要个厨师、管家或司机机器人呢？
But making science-fiction a reality is way harder than it looks.
但要想让科幻小说成为现实，并不像看上去那么容易。
Even just attempting to make robots look and move like humans is hard.
即使只是制作外表像人，行动像人的机器人也是相当困难的。
AI has gotten very smart, in specific aspects most recently in perception.
AI最近在感知方面，变得非常聪明。
Well you have speech recognition, as well as image recognition software on your cell phones.
你手机上（可能）会有语音识别和图像识别软件。
One of the pieces that AI is not still very good at is things like grasping and manipulation of physical objects.
但AI仍有一个弱点，那就是抓取和操纵实物。
Things like cooking actually requires enough fine motion skills that it will be at least need 10 years.
像烹饪的话，实际上涉及大量精细运动，而机器人要想实现至少需要10年时间。
You have now more sophisticated robots that can pick and move around and help you, more as a complement to the human capabilities rather than replacement.
现在有了更先进的机器人，可以在你身边挑选移动，帮你忙，但更多的仍是对人类能力的补充，而不是替代。
So robots can't physically cook you dinner and clean up afterward - yet.
机器人不能亲自为你下厨，并在事后进行清理打扫。
Even though most of the robots that you see online are more like this, we're not actually that far from robotic limbs that perform basic movements.
尽管你在网上看到的大多数机器人都是这样的，但实际上，我们距离机器人基本肢体运动并不遥远。
In the last 10 years, bionic prosthetic arms have come a long way.
过去10年里，仿生假肢已走过漫漫长路。
Many models have decent grip strength and individual finger articulation.
许多机器人模型现在都有良好的握力，单指关节运动灵活。
But, they're still controlled by people.
但是，他们仍然受人控制。
We don't fully understand how our muscles work.
我们尚不完全了解人类肌肉如何工作。
It's a complicated system that robotic technology can help to unravel.
这一系统相当复杂，机器人技术可以帮助我们揭开谜团。
So, that's where we are with movement, but right now that's kind of the best we've got.
所以，这就是我们关于运动的认知状态，但目前我们的水平也就仅限于此了。
And when it come to the looks department, we're really far away.
论及机器人的（类人）外表，我们真还差得很远。
Synthetic robot skin is made of things like silicone and polyurethane.
合成机器人皮肤是由硅酮和聚氨酯制成的。
It still can't bounce back or feel things.
它仍不具弹性，也无法感知事物。
As you probably know, it doesn't even look real.
你很可能知道，它甚至看起来都不像真的。
Realistic, synthetic skin will need to be sensitive to different tactile experiences like heat and cold, and no one is even close.
真正的合成皮肤应有敏锐的触感，知冷知热，目前机器人里，无人企及。
As for a humanoid robot with clever AI, a robotic skeleton, artificial limbs and flawless skin?
具有智能AI的人形机器人包括机器骨架、人造四肢及无暇皮肤，
That could take decades, or longer.
要想制造这样的机器人，可能需要几十年甚至更长的时间。
We'll have better sensors, sensor meaning cameras, smaller, higher resolution, faster, detecting what's out there, but it's not only visual.
那时我们的传感器更好，传感器就是相机，到时候它们更小，分辨率更高，反应更快，能够探测物体，不仅限于视觉。
It will be auditory. It will be tactile, fingers, hands.
可能还会有听觉、触觉、手指和手。
Maybe we're overthinking this, do we really want or need robots to be humanoid, like the replicants in Blade Runner?
也许我们想多了，我们真的想要或需要人形机器人吗，就像银翼杀手那样？
Or would we be happy with them being out of sight, like the disembodied voice on Star Trek?
有了它们，我们真的会幸福吗？他们是看不见的，就像星际迷航里的幽灵，发出声音。
If you want a household robot that walks from one room to the other, then you're probably looking for something that looks humanoid,
如果你希望家用机器人从一个房间走到另一个房间，那么你可能在寻找一些看起来像人形的东西，
but if on the other hand you're looking for a smart home, and food just appears in front of you, that's a smart home not a smart robot.
但另一方面，你在想着拥有智能家居，而食物正好出现在你面前，那是一个智能的家，而非一个智能机器人。
In terms of the challenges, they're the same, it's as to which one you prefer is really an issue.
就挑战而言，它们是一样的，而你喜欢哪一个才是真正的问题。
What it's really about is the task.
关键是任务。
Androids are great for working around the house, but several companies are already proving we don't necessarily need a humanoid robot to drive a car.
安卓机器人非常适合在家里工作，但有些公司已经证明，我们未必需要人形机器人来驾驶汽车。
Perhaps the more important obstacle to overcome isn't physicality or looks.
也许我们更需要攻克的不是肉体或外表。
If we really want intelligent robots to work for us, we're going to need them to learn to work with us.
如果我们真的想让智能机器人为我们工作，就需要让他们学着跟我们合作。
To be an effective assistant you really need to be able to read the other person's mind.
要想成为一个高效能助手，真正需要的是能读懂别人的想法。
That's the whole entire direction, that's actually happening right now.
这是整个研究方向，也是正在发生的事情。
And, in fact, there is this whole saying that the reason humans have the huge brains that we have is not to run away from the tigers in the savanna, but to be with each other.
事实上，有这样一句话：人类之所以拥有巨大的脑，并不是因为我们需要躲避草原上的老虎，而是需要彼此相处。
So what we need is to find a way to not only make robots act human but, for them to fully understand and even anticipate what's on our minds.
因此，我们需要找到一种方法，不仅使机器人像人那样运动，同时还要使机器人完全理解，甚至预期我们的想法。
And, is that what we really mean by intelligence? What is intelligence anyway? How do you test it?
这就是我们所说的智能吗？智力究竟是什么？又如何对它进行测试呢？
How do you know when a thing is intelligent?
当一样东西智能起来时，你又怎么知道呢？
Intelligence involves things like perception, things like ability to manipulate the world, emotional intelligence, social intelligence.
智力包括感知、操纵世界的能力、情商、社交智能等。
That process of how you deal with the never seen before, never heard before, never experienced before, and still you somehow know how to react is something that we don't understand.
当遇到未曾见，未曾听，未曾经历过的事物时，你不知何故知道该如何反应，这一过程我们不理解。
It'll be at least 25 years or more before we reach human level intelligence and we do reach human level intelligence, that we will have also passed the Turing test.
要想达到人类的智能水平，至少还要花上25年乃至更久。如果我们确实达到这一水平，也通过图灵测试才能证实。
Alan Turing, is often referred to as the father of modern computing.
我们经常称艾伦·图灵为现代计算机之父。
He said "a computer would deserve to be called intelligent if it could deceive a human into believing that it was human."
他说：“一台计算机如果能骗过人类，让人类相信它就是人的话，那我们就可以称之为智能。”
But anyone who's ever chatted with a bot online knows they're not great.
但任何曾与机器人有过在线聊天经历的人都知道，那些机器人还差得远呢。
They don't always understand, they're repetitive; they don't respond in the way a human would -- or they know things that no human would know off the top of their heads!
他们理解力有限，还老说重复的话，他们不会像人类那样做出（正确）反应，但或者他们知道任何人都不知道的事情！
There was this one human being who knew so much about all the arcana of Shakespeare that people essentially decided that it must be a machine because nobody with a real life would know so much about Shakespeare.
如果有一人知道关于莎士比亚的一切，那么人们基本上认为它一定是一个机器，因为现实生活中，根本没人知道莎士比亚的那么多事。
Robots are a coming, and while they may not be in humanoid form, they will be replacing the work of humans across our society.
机器人（时代）即将到来，虽然它们可能不是人形的，但它们将取代人类的某些社会工作。
Blue collar, white collar, nobody is safe.
无论是蓝领还是白领，大家都不安全。
The reality is that anything routine, whether it is cognitive or non-cognitive is going to be replaced.
事实是，无论是认知的还是非认知的，都可能被取代。
So, for example, truck driving is something that within five years it's probably going to be gone.
例如，卡车驾驶在五年内很可能会消失。
Radiologists: you got 5 years.
放射科医师：你就剩5年了。
Psychologists: we have an app for that.
心理学家：我们有个应用程序能做你的工作。
Sports and economics reporters: Too late, computers already do that job.
体育和经济记者：太晚了，计算机已经可以胜任这项工作。
Paralegals: Bye! So, how close are we to creating androids?
律师：再见！那么，制造机器人，我们还差多远？
Smart AI systems already exist and we're a decade away from building robots that move like us, though, aesthetically we've still got some major challenges to overcome.
我们已经有了人工智能系统，不过要想建造像我们那样行动的机器人尚需十年时间，但从美学角度来说，我们仍然需要克服一些重大挑战。
There's really no incentive to put all the pieces of the intelligent humanoid robot puzzle together, at least, not yet.
不过目前，要把所有智能人形机器人组装在一起，我们尚无动力，至少现在没有。
But in reality, we're not building toward a future where one type of artificially intelligent robot will exist.
但实际上，制造人工智能机器人的未来发展方向多种多样。
For better or for worse, we're heading toward a world where a whole fleet of specialized robots will be developed, each taking over its own little world.
不管是好是坏，未来可能是，人类建造出专业机器人整体舰队，每个机器人都有自己的小世界。
How close are we to getting you to subscribe? It's easy, just tap this button.
订阅节目，简单吗？很容易，只要按一下这个按钮就可以了。
So at the moment AI isn't great, but part of that might be that AI just doesn't have its own memories, so what if we gave an AI the ability to remember?
所以目前AI做得还不是很好，部分原因可能是AI没有自己的记忆，那么如果我们赋予它记忆能力呢？
To think back? Find out in this video we made, here. And thanks so much for watching Seeker.
想一想？在这里，您可以收看到我们录制的视频。非常感谢您收看《科学探索之旅》。
