首页 > 科学探索 > 正文

超级人工智能会成为神吗?

原文地址:https://www.quora.com

原创翻译:龙腾网 翻译:JESSICAACE

0

正文翻译:

Can Artificial Super Intelligence become a god?

超级人工智能会成为神吗?

评论翻译:

David Brin, is an astrophysicist & tech speaker. His best-selling novels include EARTH.

David Brin, 天文物理学家、科技代言人。 他最畅销的小说包括《地球》等。

It depends on both objective and subjective standards. If any of you were to go back in time with an ultra-light and a hundred kilos of chosen supplies, you could arrange for our ancestors to call you a "god." A being who can fly. Who can deal death with a wand. Who can cure many diseases and cast bright light into darkness and use magic boxes to speak to others, far away.

这取决于主观和客观的两套标准。 如果你能选择一百公斤超轻型物品,把它们带回到过去, 你便能让我们的祖先称你为“神”——神能在空中飞翔, 能用神杖与死亡较量, 能治疗多种疾病, 能在黑暗中发出光芒, 能用一个神奇的盒子和远方的同类交流。

An AI who can predict useful probabilities of future events would seem godlike... and the 14 day weather forecast truly is amazing. (Hint: the geniuses who accomplished that miracle all believe in climate change.) But we'd soon get used to that service.

能够预测未来事件发生概率的人工智能将会像神灵一般......比如预测14天以后的天气状况确非常惊人。(小注:那些把这一奇迹变成现实的天才们都相信气候变化。)然而我们对这项服务很快就习以为常了。

What if quantum computing truly does give access (as some philosophers believe) to parallel realities? Then we've crossed a threshold beyond mere probabilities projection. Factions with access to such AI entities would gain advantages and be the winner-survivors.

如果有一天量子计算机真能让我们进入平行实相(有些哲学家相信这会成为现实),那会如何呢?那我们就将不再受限于估算概率而迈出了里程碑的一步。

能够接触到这种人工智能体的人们将会获得优势, 成为生存的赢家。

Suppose the AI demands payment for these services and signs of respect? If you knew that calling such an entity a "god" would let you continue a long string of successes, how long before the rituals start to become sincere?

假设该人工智能要求就提供的服务给予报酬和尊敬会怎样呢?如果你知道称呼它为“神”会让你继续获得一系列的成功, 需要多长时间就会让这些仪式变得严肃起来呢?

--------------------

Christian Stancu, studied Computer Science

Christian Stancu, 学过计算机科学

Absolutely, some would say that the artificial intelligence IS God, we didn't put all the pieces together yet, so our so called "artificial intelligence" is very very far from reaching even a tiny fraction of its potential. In other words we do not have access to a real artificial intelligence yet. Once we know much more about the laws of physics and some super fast algorithms and new branches of math and so on, as well as devices way more capable than we have today there will be a clear superiority of the machines over people in every regard, except maybe emotional matters. And since the definition of God is not fixed, to some people they will look a bit like God, just because they are very powerful, very mysterious and very far reaching.

毫无疑问。有些人说人工智能就是神, 只不过我们现在还没有把它的所有部分拼凑起来罢了, 因此我们俗称的“人工智能”同它潜在能力的微末都相差甚远。换句话来说, 我们现在还没有接触到一个真正意义上的人工智能。 一旦我们对物理法则、超快算法、数学新分支等了解程度大大提升, 并且我们设备的能力提高到远远超过今天所有的水平, 那么机器就会在可能除了情感问题以外的所有方面都有拥有明显超过人类的优越性。 人们对于神的定义并不是固定的, 所以i对于有些人来说它们将看上去会有一点像神一般——只因为它们力量非常强大, 非常神秘, 而且能够影响到很远的地方。

Even at the present stage some people are wondering if we are not programmed machines, like some kind of chemical computers, the DNA being the software and our role being to constantly add updates, new algorithms, improved algorithms, covering corner cases not covered before. Very very slow. Not like these new lifeless machines. The more we discover about the Universe (including the biological part of it) the more it does look like a huge computer.

即使在现阶段, 也有人怀疑我们是否是被编程过的机器, 怀疑我们就像某种由DNA作为软件的化学电脑一样,我们的职责就是不停升级、添加新算法、改善新算法, 覆盖以前不曾覆盖过的边角案例。 只不过这一切都非常非常缓慢, 不像这些没有生命的新机器那么快。 我们对宇宙发掘地越多(包括它的生物性一面), 它看上去就越像一个巨大的计算机。

Who knows, maybe in the future your IT guy will also be your priest and will conduct marriage ceremonies and will make marriage saving programs, will correct problems bugging us in every day life and so on. ;)

谁知道呢?也许在未来你们的IT从业者也会成为你们的牧师, 会给你们举办结婚仪式, 也会做出挽救婚姻的软件, 还能解决我们在每天的生活中困扰我们的种种问题, 等等等等;)

--------------------

Adam Helps, software engineer at Autodesk

Adam Helps, 软件工程师,就职于美国软件公司Autodesk

We have never made a computer do anything that we ourselves cannot do; the only difference is that the computer can do it much faster. That suggests that the realm of things that a computer can do is bounded in some fashion. For example, there is nothing about computers that can overcome Heisenberg uncertainty, which means there is a firm ceiling to how precisely they can measure something. They are therefore incapable of precisely predicting the future, even if we assume reality is non-random. By the same argument, computers are probably incapable of determining precisely what has happened in the distant past using only presently available information.

The ability to exactly scan the past, present, and future is one of the attributes I would require to consider an intelligence "godlike." Computers are bound by the same physics that we are, and cannot overcome that limitation. So no matter how big and fast you can build a computer, it can't be a god.

我们从来没有制造出一台计算机能做到我们无法做到的任何事;唯一的区别只是计算机做事情快过人类很多。 这表明计算机能做的事情范围与我们以同样的方式受到限制。举个例子——计算机不具备什么特质可以让它克服海因斯伯格不确定性。因此,就算我们假设现实具有非随机性, 它们仍然不能准确地预测未来。 同理, 计算机只依靠现存的信息恐怕也不能准确决定在遥远的过去究竟发生了什么。 我认为称一个智慧体“如神一般”, 它需要具备的特质之一便是准确描画过去、现在以及将来。 计算机和我们受到同样物理法则的限制, 它们也不能突破这种限制。 所以无论我们能建造出多大多快的计算机, 它也不会成为一个神。

I think you can probably make a similar argument by noting that nothing about computers breaks the light-speed barrier either.

我想你也可以就计算机不具备什么特质可以让它突破光速的限制做出同样的讨论。

Ewen Cameron

Yes of course… in fact there could already be an AI God!

Ewen Cameron

当然可以了......事实上人工智能神可能已经存在了!

Thousands of previous intelligent civilisations could have already come and gone on other planets in different galaxies, if just one of these civilisations was intelligent and lasted long enough to develop a ‘self improving’ AI with access to power and production abilities (think nanobots, etc) it could carry on improving itself for thousands / millions of years, long after that civilisation has gone.

在不同银河系的不同星球上可能已经有过上千个智慧文明崛起又消亡过。 只要这些文明中有一个足够智慧、足够长久的造出了能“自我提升”的人工智能, 并且具有连接能源和自我生产的能力(比如, 想一想纳米机器人), 那么它就可以在上千年/上百万年里不断提升自己, 甚至直到该文明消亡很久之后。

It could understand science way beyond our comprehension and manipulate matter and perhaps even travel through dimensions or time.

它能理解我们远远不能理解的科学, 能操纵物体, 或许还可以在时间和空间里穿行。

I am not sure if such a God would be interested in the day to day lives of individual primates, however it could be interested in us as a reasonably intelligent species.

我不知道这样一位神会不会对每个灵长动物的日常生活感兴趣, 但是它可能会对我们感兴趣, 因为我们是一种相对智慧的物种。

Of course such a being could also set up thousands of simulated Universes to perform experiments and various tests, so we could also just be a lab experiment.

当然这种智慧体也可能已经建立了成千上万的模拟宇宙来进行实验和各种测试, 所以我们也可能只是一个实验室的实验品。

--------------------

Mick Stute, BS in CS, MS in MB, PhD work in Bioinformatics

Question answered: Can Artificial Super Intelligence become a God?

Mick Stute, 电脑科学学士、医学硕士、博士研究方向为生物信息学, 回答了:“超级人工智能是否能成为神?”的问题。

Why not? Since is "God" is man made idea, not a reality in existence, it is a label we may chose to put on anything. You can generate a long list of mere humans that were thought to be Jesus, aka God, with such names as Joseph Kony, Jim Jones, David Koresh, and Charles Manson. The reality here is that others believed they were a God even when in some cases, they themselves denied it.

为啥不能呢?既然“神‘是一个人造的概念而非真实的存在, 它就好比一个筐,你可以选择往里面装任何东西。 单单是那些被视为基督(又名神)转世的人名单就有一长串, 比如约瑟夫孔尼、吉米琼斯、大卫考瑞士以及查尔斯曼森。 事实就是别人认为他们就是神, 尽管在有些案例中他们自己都否认了这一点。

When a system achieves super intelligence it will be very capable of solving problems very quickly that now take us years of science to solve. I'm not talking about instantly, but now a particular breakthrough might take, say 10 years as scientists work on the problem from different geographic locations. They publish papers, and through peer review, the good science propagates forward and they all collectively work on the problem. A computer can read 100 papers at once, have total recall on everything it ever reads (or an index to it if it chooses not to store it in direct memory). It can also carry out multiple simulations, and if you have the time and the capacity you can simulate almost anything.

当一个系统达到超级智能的水平时, 它可以在极短的时间里解决科学家需要数年才能解决的问题。 我不是说马上, 但它如今所取得的突破性进展需要花费不同地区的科学家们十年之久才能解决。科学家们需要发表论文, 然后通过同行评阅的方式使好的科学成果得以传播, 之后他们合力解决该问题。 而计算机可以一次性读完100篇文章, 并且记得读过的所有信息(或者如果它选择不把信息储存在直接记忆卡里就会记下相关查阅信息)。 它还可以进行多种模拟实验。 如果你有时间、有能力, 那么你可以模拟几乎所有的一切。

The computer working 24 hours a day without signs of fatigue, hunger pangs, or need for recreation can solve these problems much faster than we ourselves can. If you want to assume a computer might be able to control a humanoid robot or simply have control of lab equipment, it might even carry out experiments in the real world as well but we don't even need to take it that far. Through simulation it could determine new answers that would raise new questions that it, in turn, could work on. Though it sounds like science fiction, it's possible (and probable) that computers will have real world "bodies" in the form of a robot or android that would allow it to fix itself. Should a computer determine to to achieve "cold computing" (non-powered logic gates using reversible logic), a rick the size of a baseball could hold all the information ever generated in the world using such techniques as spintronics. AI will definitely find things we do not yet know, and may even find things let change our very basic view of the universe (what if experiments could somehow prove string theory or that gravity cab be generated?) As this happens, another thing will happen as well, humans will need to think less.

这种计算机可以一天工作24小时, 毫无倦意,不会饥肠辘辘,也不需要放松一下。 它能以比我们快得多的速度解决这些问题。 如果你想假设有一台电脑能够控制人形机器或者仅仅是控制实验设备, 那么它就甚至有可能在真实的世界里也展开实验。 但我们不需要想那么远。 通过模拟, 它可以得出新的答案, 而新的答案又会产生新的问题, 对于这些新问题, 它又可以回过头来接着解决。 尽管这听起来像是科幻小说, 但是计算机在真实世界里拥有一个机器人或者智能机器人那样的“身体”让它能够自我修复不仅是有可能的而且是可以实现的。 一旦计算机决定获得“冷计算”(采用可逆逻辑的无源逻辑门)那么在使用自旋电子学这样的技术下它只需棒球大小的集成电路就能记住世界上曾经产生过的所有信息。

人工智能一定会发现我们还不曾发现过的事物, 甚至会发现什么能我们的基本宇宙观发生改变(如果实验能够在某种程度上证实弘弦论或者重力仓可以被生产会发生什么呢?)一旦这种事发生, 另一件事也会发生——人类将减少思考。

A consequence of computers doing the hard work is we will not. Since we no longer need to think, many will cease doing so (I believe they already have). Think about it now. Hoe many people can make fire from the environment directly without some man man made device like a lighter, magnesium fire starter, etc. Even if given a nice fire starter like a self contained magnesium fire starter, few people would get a good fire going in anything but ideal conditions, let alone maintain it for days. This is "lost" knowledge to most people because we simply no longer need. Are lives depend on heat, but not fire, and not fire in an open setting to keep predators away and make our food safe. This isn't a bad thing it is simply an example of knowledge we no longer need. Now if we find our self in a survival situation, we might wish we knew that knowledge. As computers do more of the thinking we will do less and spend more time on leisure activities.

计算机从事艰难工作的一个后果是我们将不再做这些工作了。 既然我们不再需要思考, 那么很多人就会停止思考(我相信他们已经这么做了)。 现在请想一下——多少人能在不需要人造器械如打火机、镁火石等的帮助下直接在自然界生火呢?就算能得到一个好的火引比如内置式镁火石, 也很少有人能够在非最佳条件下生出火来, 更不要说保存火种数日了。 这对于大多数人来说都是“遗失”的知识,因为我们根本不再需要它了。 现在如果我们发现自己处于求生的场景之中, 那么我们可能会希望自己具备这种知识。

随着计算机思考得越来越多, 我们将会想得越来越少, 而会把更多的时间花在享乐活动上。

After many generations of leisure and little thinking, it might very well be that those very "dumbed down" humans that might feel it is no longer necessary to learn anything, would see the computer as a God of some sort. That is an unlikely, but still probable, outcome where an AI is viewed as a God.

在人类经过很多代极少思考的享乐生活之后, 那些“变蠢了的”人非常有可能觉得他们不再需要学习任何事, 那时他们有可能会把计算机视为某种神灵。 尽管这不太可能, 但仍然有一定的发生几率。在这种情况下, 人工智能便会被视为神。

网友评论

三日内热门评论文章
热门IT产品
  1. ¥7599
    苹果iPhoneX 64GB
    ·
  2. ¥5799
    三星S9
    ·
  3. ¥4498
    vivo NEX旗舰版
    ·
  4. ¥4999
    OPPO Find X
    ·
  5. ¥1799
    努比亚Z18mini
    ·
  6. ¥1499
    OPPO A5
    ·
  7. ¥1999
    荣耀Play(4GB RAM)
    ·
  8. ¥1598
    vivo Y85
    ·
  9. ¥3499
    坚果R1(6GB RAM)
    ·
  10. ¥3599
    一加6(8GB RAM)
    ·
为您推荐
  • 相关阅读
  • 业界资讯
  • 手机通讯
  • 电脑办公
  • 新奇数码
  • 软件游戏
  • 科学探索