声明: 本站全部内容源自互联网,不进行任何盈利行为

仅做 整合 / 美化 处理

首页: https://dream-plan.cn

【TED】无需畏惧超级人工智能

 

When I was a kid, I was the quintessential nerd. 在我还是孩子的时候,我是一个典型的书呆子。 I think some of you were, too. 我猜你们的一部分人和我一样。 (Laughter) (笑声) And you, sir, who laughed the loudest, you probably still are. 还有你,那位先生,笑得最大声,说不定你现在还是呢。 (Laughter) (笑声) I grew up in a small town in the dusty plains of north Texas, 我成长在德克萨斯州北部荒凉平原上的一个小镇。 the son of a sheriff who was the son of a pastor. 我的爸爸是警长,爷爷是一位牧师, Getting into trouble was not an option. 所以自找麻烦从来不是一个好的选项。 And so I started reading calculus books for fun. 所以我开始看有关微积分的书来打发时间。 (Laughter) (笑声) You did, too. 你也是这样。 That led me to building a laser and a computer and model rockets, 于是我学着制作了一个激光器,一台电脑,和一个火箭模型, and that led me to making rocket fuel in my bedroom. 并且在自己的卧室制取火箭燃料。 Now, in scientific terms, 现在,从科学的角度而言, we call this a very bad idea. 我们把这个叫做一个糟糕透顶的主意。 (Laughter) (笑声) Around that same time, 在差不多同一时间, Stanley Kubrick's "2001: A Space Odyssey" came to the theaters, Stanley Kubrick的 “2001:太空漫游”上映了, and my life was forever changed. 我的生活从此改变。 I loved everything about that movie, 我喜欢关于那部电影的一切, especially the HAL 9000. 特别是 HAL 9000。 Now, HAL was a sentient computer HAL是一台有情感的电脑, designed to guide the Discovery spacecraft from the Earth to Jupiter. 为引导“发现一号”飞船从地球前往木星而设计出来。 HAL was also a flawed character, HAL 也是一个有缺陷的角色, for in the end he chose to value the mission over human life. 因为在结尾他将任务的价值置于生命之上。 Now, HAL was a fictional character, HAL 是一个虚构的角色, but nonetheless he speaks to our fears, 但尽管如此,他挑战了我们的恐惧, our fears of being subjugated by some unfeeling, artificial intelligence who is indifferent to our humanity. 被一些冷漠无情的人工智能(AI)所统治的恐惧。 I believe that such fears are unfounded. 但我相信这些恐惧只是杞人忧天。 Indeed, we stand at a remarkable time in human history, 的确,我们正处于人类历史上 where, driven by refusal to accept the limits of our bodies and our minds, 一个值得铭记的时间点,不甘被自身肉体和头脑所局限, we are building machines of exquisite, beautiful complexity and grace that will extend the human experience in ways beyond our imagining. 我们正在制造 那些可以通过我们无法想象的方式 来拓展人类体验的机器,它们精美,复杂,而且优雅。 After a career that led me from the Air Force Academy to Space Command to now, 我在空军学院和航天司令部工作过, I became a systems engineer, 现在是一个系统工程师。 and recently I was drawn into an engineering problem associated with NASA's mission to Mars. 最近我碰到了一个和NASA火星任务相关的工程问题。 Now, in space flights to the Moon, 当前,前往月球的宇宙飞行, we can rely upon mission control in Houston 我们可以依靠休斯顿的指挥中心 to watch over all aspects of a flight. 来密切关注飞行的各个方面。 However, Mars is 200 times further away, 但是,火星相比而言多出了200倍的距离, and as a result it takes on average 13 minutes for a signal to travel from the Earth to Mars. 这使得一个信号从地球到火星平均要花费13分钟。 If there's trouble, there's not enough time. 如果有了麻烦,我们并没有足够的时间来解决。 And so a reasonable engineering solution 所以一个可靠的工程解决方案 calls for us to put mission control 促使我们把一个指挥中枢 inside the walls of the Orion spacecraft. 放在“猎户星”飞船之中。 Another fascinating idea in the mission profile 在任务档案中的另一个创意 places humanoid robots on the surface of Mars before the humans themselves arrive, 是在人类抵达火星之前,把人形机器人先一步放在火星表面, first to build facilities 它们可以先建造据点, and later to serve as collaborative members of the science team. 然后作为科学团队的合作伙伴驻扎。 Now, as I looked at this from an engineering perspective, 当我从工程师的角度看待这个想法, it became very clear to me that what I needed to architect was a smart, collaborative, socially intelligent artificial intelligence. 对于建造一个聪明,懂得合作,擅长社交的人工智能的需求是十分明显的。 In other words, I needed to build something very much like a HAL 换句话说,我需要建造一个和HAL一样, but without the homicidal tendencies. 但是没有谋杀倾向的机器。 (Laughter) (笑声) Let's pause for a moment. 待会儿再回到这个话题。 Is it really possible to build an artificial intelligence like that? 真的有可能打造一个类似的人工智能吗? Actually, it is. 是的,当然可以。 In many ways, 在许多方面, this is a hard engineering problem with elements of AI, 一个困难的工程问题来自于AI的各个零件, not some wet hair ball of an AI problem that needs to be engineered. 而不是什么琐碎的AI问题。 To paraphrase Alan Turing, 借用Alan Turing的话来说, I'm not interested in building a sentient machine. 我没有兴趣建造一台有情感的机器。 I'm not building a HAL. 我不是要建造HAL。 All I'm after is a simple brain, 我只想要一个简单的大脑, something that offers the illusion of intelligence. 一个能让你以为它有智能的东西。 The art and the science of computing have come a long way since HAL was onscreen, 自从HAL登上荧幕,关于编程的技术和艺术已经发展了许多, and I'd imagine if his inventor Dr. Chandra were here today, 我能想象如果它的发明者Chandra博士今天在这里的话, he'd have a whole lot of questions for us. 他会有很多的问题问我们。 Is it really possible for us to take a system of millions upon millions of devices, 我们真的可能用一个连接了无数设备的系统, to read in their data streams, 通过读取数据流, to predict their failures and act in advance? 来预测它们的失败并提前行动吗? Yes. 是的。 Can we build systems that converse with humans in natural language? 我们能建造一个和人类用语言交流的系统吗? Yes. 能。 Can we build systems that recognize objects, identify emotions, 我们能够打造一个能辨识目标,鉴别情绪, emote themselves, play games and even read lips? 表现自身情感,打游戏,甚至读唇的系统吗? Yes. 可以。 Can we build a system that sets goals, 我们可以建造一个能设定目标, that carries out plans against those goals and learns along the way? 通过各种达成目的的方法来学习的系统吗? Yes. 也可以。 Can we build systems that have a theory of mind? 我们可以建造一个类似人脑的系统吗? This we are learning to do. 这是我们正在努力做的。 Can we build systems that have an ethical and moral foundation? 我们可以建造一个有道德和感情基础的系统吗? This we must learn how to do. 这是我们必须要学习的。 So let's accept for a moment 总而言之, that it's possible to build such an artificial intelligence for this kind of mission and others. 建造一个类似的用于这类任务的人工智能是可行的。 The next question you must ask yourself is, 另一个你必须扪心自问的问题是, should we fear it? 我们应该害怕它吗? Now, every new technology 毕竟,每一种新技术 brings with it some measure of trepidation. 都给我们带来某种程度的不安。 When we first saw cars, 我们第一次看见汽车的时候, people lamented that we would see the destruction of the family. 人们悲叹我们会看到家庭的毁灭。 When we first saw telephones come in, 我们第一次看见电话的时候, people were worried it would destroy all civil conversation. 人们担忧这会破坏所有的文明交流。 At a point in time we saw the written word become pervasive, 在某个时间点我们看到书写文字的蔓延, people thought we would lose our ability to memorize. 人们认为我们会丧失记忆的能力。 These things are all true to a degree, 这些在某个程度上是合理的, but it's also the case that these technologies 但也正是这些技术 brought to us things that extended the human experience in some profound ways. 给人类的生活在某些方面带来了前所未有的体验。 So let's take this a little further. 我们再稍稍扩展一下。 I do not fear the creation of an AI like this, 我并不畏惧这类人工智能的诞生, because it will eventually embody some of our values. 因为它最终会融入我们的部分价值观。 Consider this: building a cognitive system is fundamentally different 想想这个:建造一个认知系统 than building a traditional software-intensive system of the past. 与建造一个以传统的软件为主的系统有着本质的不同。 We don't program them. We teach them. 我们不编译它们。我们教导它们。 In order to teach a system how to recognize flowers, 为了教会系统如何识别花朵, I show it thousands of flowers of the kinds I like. 我给它看了上千种我喜欢的花。 In order to teach a system how to play a game -- 为了教会系统怎么打游戏—— Well, I would. You would, too. 当然,我会。你们也会。 I like flowers. Come on. 我喜欢花。这没什么。 To teach a system how to play a game like Go, 为了教会系统如何玩像围棋一样的游戏, I'd have it play thousands of games of Go, 我玩了许多围棋的游戏, but in the process I also teach it 但是在这个过程中 how to discern a good game from a bad game. 我也教会它如何分别差游戏和好游戏。 If I want to create an artificially intelligent legal assistant, 如果我想要一个人工智能法律助手, I will teach it some corpus of law 我会给它一些法律文集, but at the same time I am fusing with it 但同时我会将怜悯和正义 the sense of mercy and justice that is part of that law. 也是法律的一部分这种观点融入其中。 In scientific terms, this is what we call ground truth, 用一个术语来解释,就是我们所说的真相, and here's the important point: 而关键在于: in producing these machines, 为了制造这些机器, we are therefore teaching them a sense of our values. 我们正教给它们我们的价值观。 To that end, I trust an artificial intelligence 正因如此,我相信一个人工智能 the same, if not more, as a human who is well-trained. 绝不逊于一个经过良好训练的人类。 But, you may ask, 但是,你或许会问, what about rogue agents, 要是流氓组织, some well-funded nongovernment organization? 和资金充沛的无政府组织也在利用它们呢? I do not fear an artificial intelligence in the hand of a lone wolf. 我并不害怕独狼掌控的人工智能。 Clearly, we cannot protect ourselves against all random acts of violence, 很明显,我们无法从随机的暴力行为中保护自己, but the reality is such a system 但是现实是,制造这样一个系统 requires substantial training and subtle training 超越了个人所拥有资源的极限, far beyond the resources of an individual. 因为它需要踏实细致的训练和培养。 And furthermore, 还有, it's far more than just injecting an internet virus to the world, 这远比向世界散播一个网络病毒, where you push a button, all of a sudden it's in a million places 比如你按下一个按钮,瞬间全世界都被感染, and laptops start blowing up all over the place. 并且在各处的笔记本电脑中开始爆发来的复杂。 Now, these kinds of substances are much larger, 这类东西正在越来越强大, and we'll certainly see them coming. 我们必然会看到它们的来临。 Do I fear that such an artificial intelligence might threaten all of humanity? 我会害怕一个有可能威胁所有人类的人工智能吗? If you look at movies such as "The Matrix," "Metropolis," 如果你看过《黑客帝国》,《大都会》, “The Terminator,” shows such as “Westworld,” 《终结者》,或者 《西部世界》这类电视剧, they all speak of this kind of fear. 它们都在表达这种恐惧。 Indeed, in the book "Superintelligence" by the philosopher Nick Bostrom, 的确,在哲学家Nick Bostrom 写的《超级智能》中, he picks up on this theme 他选择了这个主题, and observes that a superintelligence might not only be dangerous, 并且观察到超级智能不仅仅危险, it could represent an existential threat to all of humanity. 它还对所有人类的存在造成了威胁。 Dr. Bostrom's basic argument Bostrom博士的基础观点认为, is that such systems will eventually have such an insatiable thirst for information 这样的系统迟早会产生对信息的无止境渴求, that they will perhaps learn how to learn 也许它们会开始学习, and eventually discover that they may have goals that are contrary to human needs. 并且最终发现它们的目的和人类的需求背道而驰。 Dr. Bostrom has a number of followers. Bostrom博士有一群粉丝。 He is supported by people such as Elon Musk and Stephen Hawking. Elon Musk和Stephen Hawking也支持他。 With all due respect to these brilliant minds, 虽然要向这些伟大的头脑致以崇高的敬意, I believe that they are fundamentally wrong. 但我还是相信他们从一开始就错了。 Now, there are a lot of pieces of Dr. Bostrom's argument to unpack, Bostrom博士的观点有很多地方可以细细体会, and I don't have time to unpack them all, 但现在我没有时间一一解读, but very briefly, consider this: 简要而言,请考虑这句话: super knowing is very different than super doing. 全知并非全能。 HAL was a threat to the Discovery crew HAL成为了对发现一号成员的威胁, only insofar as HAL commanded all aspects of the Discovery. 只是因为它控制了发现一号的各个方面。 So it would have to be with a superintelligence. 正因如此它才需要是一个人工智能。 It would have to have dominion over all of our world. 它需要对于我们世界的完全掌控。 This is the stuff of Skynet from the movie “The Terminator” 这就是《终结者》中的天网, in which we had a superintelligence 一个控制了人们意志, that commanded human will, that directed every device that was in every corner of the world. 控制了世界各处各个机器的超级智能。 Practically speaking, 说实在的, it ain't gonna happen. 这完全是杞人忧天。 We are not building AIs that control the weather, 我们不是在制造可以控制天气, that direct the tides, 引导潮水, that command us capricious, chaotic humans. 指挥我们这些多变,混乱的人类的人工智能。 And furthermore, if such an artificial intelligence existed, 另外,即使这类人工智能存在, it would have to compete with human economies, 它需要和人类的经济竞争, and thereby compete for resources with us. 进而和我们拥有的资源竞争。 And in the end -- 最后—— don't tell Siri this -- 不要告诉Siri—— we can always unplug them. 我们可以随时拔掉电源。 (Laughter) (笑声) We are on an incredible journey of coevolution with our machines. 我们正处于和机器共同演化的奇妙旅途之中。 The humans we are today are not the humans we will be then. 未来的人类将与今天的人类大相径庭。 To worry now about the rise of a superintelligence 当前对人工智能崛起的担忧, is in many ways a dangerous distraction 从各方面来说都是一个危险的错误指引, because the rise of computing itself 因为电脑的崛起 brings to us a number of human and societal issues to which we must now attend. 带给了我们许多必须参与的关乎人类和社会的问题。 How shall I best organize society 我应该如何管理社会 when the need for human labor diminishes? 来应对人类劳工需求量的降低? How can I bring understanding and education throughout the globe 我应该怎样在进行全球化交流和教育的同时, and still respect our differences? 依旧尊重彼此的不同? How might I extend and enhance human life through cognitive healthcare? 我应该如何通过可认知医疗来延长并强化人类的生命? How might I use computing to help take us to the stars? 我应该如何通过计算机来帮助我们前往其他星球? And that's the exciting thing. 这些都很令人兴奋。 The opportunities to use computing to advance the human experience 通过计算机来升级人类体验的机会 are within our reach, 就在我们手中, here and now, 就在此时此刻, and we are just beginning. 我们的旅途才刚刚开始。 Thank you very much. 谢谢大家。 (Applause) (掌声)

萌ICP备20223985号