大卫·多伊奇：知识创造与人类种族，第一部分

我的目标不是再做一期与大卫·多伊奇相同的播客。

My goal would be not to do yet another podcast with David Deutsch.

这类内容已经够多了。

There are plenty of those.

我希望能梳理出一些非常反直觉的见解，将它们以经典形式记录下来，让后世受益，确保这些思想不会遗失。

I would love to tease out some of the very counterintuitive learnings, put them down canonically in such a way that future generations can benefit from them, and make sure that none of this is lost.

你的著作对我产生了难以置信的影响。

Your work has been incredibly influential for me.

无论走到哪里，我随身都带着《无限的开端》或《现实的构造》的副本。

I am always carrying a copy of Beginning of Infinity or Fabric of Reality with me wherever I go.

两年过去了，我仍在反复阅读这些书，试图将它们融入我的世界观，每天都能从中获得新的领悟。

I'm still reading these same books after two years, trying to absorb them into my worldview and I learn something new from them every day.

书中有许多反直觉的观点。

There's a lot of counter intuitive things in there.

你戳穿了许多神圣教条和陈词滥调。

There are a lot of sacred dogmas and shibbolethes that you're skewering.

有时你只是随口一句话，就让我花上数周才能完全理解。

Sometimes you do it in passing with a single sentence that takes me weeks to unpack properly.

这段录音不是为哲学家准备的，也不是为物理学家准备的。

This recording is not for the philosophers, it's not for the physicists.

这是面向普通大众的，我们想向他们介绍乐观主义原则、无限开端理论、可持续发展的真谛，以及拟人化错觉。

This is for the layman, the average person, and we want to introduce them to the principles of optimism, the beginning of infinity, what sustainability really means, about anthropomorphic delusions.

比如，你推翻了归纳法作为形成新科学理论的途径。

As an example, you overturn induction as a way of forming new scientific theories.

这种认为重复观察导致新知识产生的观念完全不符合事实。

That's this idea that repeated observation is what leads you to the creation of new knowledge and that's not the case at all.

这显然源自波普尔的理论，但你在其基础上进行了发展。

This obviously came from Popper, but you built upon it.

你谈到人类如何与众不同，知识创造是进化中仅发生在人类大脑中的非凡现象——就目前所知而言。

You talk about how humans are very different and very exceptional and knowledge creation is exceptional thing that only happens in evolution and the human brain as far as we know.

你还论述地球并非那个适宜居住、脆弱的'太空船地球'生物圈，而是我们通过工程手段建造来维持生存的环境。

And you talk about how the earth is not this hospitable, fragile, spaceship earth biome that supports us, but rather it's something that we engineer and we build to sustain us.

我总是建议人们先读《无限起源》的前三章，它们通俗易懂，却颠覆了人们在基础推理中视为理所当然的核心信条——其颠覆性超过我见过的几乎所有书籍。

I always recommend to people start with the first three chapters of Beginning of Infinity because they're easy to understand, but they overturn more central dogmas that people are taking for granted in base reasoning than almost any other book I've ever seen.

我认为有必要向听众指出，你的哲学体系并非基于任意公理形成的世界观。

I think it's important to point to listeners that your philosophy isn't just some arbitrary set of axioms based on which you view the world.

我更愿将其视为由优质解释和实验证据支撑的晶体结构，形成自洽的认知体系，它运作于你在《真实之网》中论述的四个领域交汇处：认识论、计算学、物理学和进化论。

I think of it as a crystalline structure held together by good explanations and experimental evidence that then forms a self consistent view of how things work, and it operates at the intersection of these four strands that you talk about in the fabric of reality, epistemology, computation, physics and evolution.

让我们深入探讨人类这个话题。

Let's get into humans.

传统模型认为：从鱼类开始，到蝌蚪、青蛙、某种猿猴、再到直立弓背的生物，人类只是这个动物序列的延续。

So there's a classic model, you start with a fish, and then it turns to a tadpole and then a frog and then some kind of monkey and then an upright hunched over creature and a human is just this progression along all the animals.

但在你的理解和解释中，这里发生了根本性的质变。

But in your understanding, in your explanation, there's something fundamentally different that happens.

你在一个精彩视频中阐述过这个观点，我强烈推荐大家搜索观看。

And you talked about this in a great video, which I encourage everybody to look up.

视频标题是《做着遥远类星体梦的化学浮沫》。

It's titled chemical scum that dream of distant quasars.

人类是什么？

What are humans?

他们有何独特之处？

How are they unique?

他们又为何卓越非凡？

And how are they exceptional?

我们该如何看待人类相对于地球上其他物种的地位？

And how should we think of the human species relative to the other species that are on this planet?

每种动物都有其独特之处。

Every animal is exceptional in some way.

否则我们也不会将不同物种区分开来。

Otherwise, we wouldn't call different species different species.

有飞得最快的鸟，也有飞得最高的鸟，诸如此类。

There's the bird that can fly faster than any other bird, and there's a bird that can fly higher than any other one, and so on.

直觉告诉我们，人类的独特性比其他物种的所有特性都更为重要。

It's intuitively obvious that we are unique in some way that's more important than all those other ways.

正如我在《无穷的开始》开篇所言，全球许多科学实验室里都备有香槟。

As I say in the beginning of infinity, in many scientific laboratories around the world, there is a champagne bottle.

那些酒瓶和冰箱都是实体物件。

That bottle and that fridge are physical objects.

参与其中的人也是物理实体。

The people involved are physical objects.

他们都遵循物理定律。

They all obey the laws of physics.

然而，为了理解人类将香槟长期存放在冰箱中的行为，我在想象外星人观察人类时，他们必须理解人类试图达成的目标以及能否实现。

And yet in order to understand the behavior of humans in regard to champagne bottles stored for long periods in fridges I'm thinking of aliens looking at humans they have to understand what those humans are trying to achieve and whether they will or won't achieve it.

换句话说，如果你是一个俯视地球、观察那里发生的一切并试图解释的外星人，为了解释地球上发生的所有事情，假设这些外星人与我们截然不同，我们身上没有任何他们熟悉的东西。

In other words, if you were an alien that was looking down on the earth and seeing what's happening there and was trying to explain it, in order to explain everything that happens on earth and let's suppose that these aliens are so different from us, there's nothing familiar about us.

要理解地球上发生的事情，他们需要知晓一切，毫不夸张地说。

In order to understand stuff that happens on earth, they would need to know everything, literally.

例如广义相对论，因为他们需要用它来解释为什么一个叫爱因斯坦的猴子被带到瑞典并获得了金子。

For example, general relativity because they need that to explain why this one monkey Einstein was taken to Sweden and given some gold.

如果你想解释这点，就必须援引广义相对论。

If you want to explain that, you've got to invoke general relativity.

有些人因发明了点数学而获得菲尔兹奖。

Some people get the Fields Medal for inventing a bit of mathematics.

要理解那个人为何获得菲尔兹奖，他们必须懂数学。

To understand why that person won the Fields Medal, they'd have to understand mathematics.

而这还远没有尽头！

And there's no end to this!

他们必须理解全部科学、整个物理学，甚至整个哲学和道德体系。

They have to understand the whole of science, the whole of physics, even the whole of philosophy, morality.

其他任何动物都不具备这种特性。

This is not true of any other animal.

其他任何物理对象也不具备这种特性。

It's not true of any other physical object.

对于所有其他物理对象，即便是类星体等重要天体，你只需要物理学定律的极小片段就能详尽理解它们的行为。

For all other physical objects, even really important ones like quasars and so on, you only need a tiny sliver of the laws of physics in order to understand their behavior in any kind of detail.

换句话说，要充分理解人类，就必须充分理解一切。

In other words, to understand humans sufficiently well you must understand everything sufficiently well.

而人类是我们已知宇宙中唯一符合这一标准的物理系统。

And humans are the only remaining physical systems that we know of in the universe of which that is true.

从这个意义上说，其他一切事物都无足轻重。

Everything else is really inconsequential in that sense.

你对知识有个绝妙的定义——关于知识如何在环境中自我延续，这是大多数人甚至未曾尝试探讨的。

You have a beautiful definition of knowledge, which most people don't even try and tackle, about how knowledge perpetuates itself in the environment.

你举了一些非常精彩的例子。

There were some really good examples you gave.

其中一个与基因相关。

One was around genes.

成功高度适应的基因蕴含大量知识，因此它们能促使自身复制，因为它们是生存的赢家。

Successful highly adapted genes contain a lot of knowledge, so they cause themselves to be replicated because they're survivors.

同理，知识本身也是生存者——如果你传授我计算机的制造知识，这项极其有用的知识会促使我制造更多计算机，从而使知识得以传承。

And the same way, knowledge itself is a survivor in that if you transmit to me the knowledge of how to build a computer, it's an incredibly useful thing, so I'm going build more and more computers and that knowledge will be passed on.

你在此反复强调的核心观点是：若要理解物理宇宙，就必须理解知识，因为正是知识随着时间推移逐渐占据主导，并比其他任何事物都更深刻地改变着宇宙。

And your underlying point that you repeated here was, if you want to understand the physical universe, you have to understand knowledge because it is the thing that over time takes over and changes more and more the universe than almost anything else.

你必须理解其背后的所有解释。

You have to understand all the explanations behind it.

你不能仅用粒子碰撞解释一切，因为这样等于什么都没解释。

You can't just say particle collisions because that explains everything, so it explains nothing.

这不是一个具有操作意义的层面。

It's not a useful level to operate at.

因此，创造知识的事物在宇宙中具有独特的影响力。

Therefore, the things that create knowledge are uniquely influential in the universe.

据我们所知，目前只有两种系统能够创造知识。

And as far as we know, there are only two systems that create knowledge.

一种是进化，另一种是人类。

There's evolution, and there's humans.

但即便是这两种知识创造形式之间也存在差异，不是吗？进化与人类之间的区别？

But there's a difference even between these two forms of knowledge creation, aren't there, between evolution and between humans?

是的。

Yes.

我曾论证过，人类创造知识的方式是终极的，不存在比这更强大的方式。

I have argued that the human way of creating knowledge is the ultimate one, that there aren't any more powerful ones than that.

这是反对超自然存在的论点。

This is the argument against the supernatural.

假设存在比我们更强大的知识创造形式，就等于诉诸超自然，这始终是个糟糕的解释。

Assuming that there is a form of knowledge creation that's more powerful than our one is equivalent to invoking supernatural, which is therefore a bad explanation, as invoking the supernatural always is.

生物进化与人类创造性思维的区别在于，生物进化本质上受限于其范围，因为它缺乏预见能力。

The difference between biological evolution and human creative thought is that biological evolution is inherently limited in its range, and that is because biological evolution has no foresight.

它无法发现问题并推测解决方案。

It can't see a problem and conjecture a solution.

每当生物进化产生某个问题的解决方案时，自然选择甚至都还未开始运作。

Whenever biological evolution produces a solution to something, it's always before natural selection has even begun.

这就是查尔斯·达尔文的洞见。

This is Charles Darwin's insight.

这就是查尔斯·达尔文的进化论与之前存在了一个多世纪的其他进化理论（包括达尔文的祖父和拉马克的理论）之间的区别。

This is the difference between Charles Darwin's theory of evolution and the other theories of evolution that had been around for a century or more before that, including Charles Darwin's grandfather and Lamarck.

他们未能理解的是，进化中知识的创造在此之前就已经开始了。

The thing they didn't get is that the creation of knowledge in evolution begins before.

这意味着生物进化无法触及那些无法通过连续改进达到的领域，其中每一步改进都能让一个可存活的生物体存在。

That means that biological evolution can't reach places that are not reachable by successive improvements, that each of which allows a viable organism to exist.

神创论者认为生物进化实际上已经达到了无法通过渐进步骤实现的事物，而每一步骤本身都是可存活的生物体。

Creationists say that biological evolution has in fact reached things that are not reachable by incremental steps, each of which is a viable organism.

他们在事实上是错误的。

They're factually mistaken.

但他们心中的概念是一个能够想象不存在之物的造物主，能够创造出并非由一系列可存活事物累积而成的理念。

But the thing which they have in mind is the idea of a creator who can imagine things that don't exist and who can create an idea that is not the culmination of a whole load of viable things.

一个有思维的个体可以创造出由一系列不可行事物累积而成的产物。

A thinking being can create something that's a culmination of a whole load of non viable things.

在所有存在过的数以亿计的物种中，没有一个物种曾生起过篝火，尽管许多物种本可以从具备生火的基因能力中受益。

Out of all the billions and billions of species that have ever existed, none of them has ever made a campfire, even though many of them would have been helped by having the genetic capacity to make campfires.

这种现象未在生物圈发生的原因在于，不存在所谓'部分功能性的篝火'。

The reason it didn't happen in the biosphere is that there is no such thing as making a partially functional campfire.

然而有些功能确实可以逐步进化，比如制造热水——投弹甲虫就会向敌人喷射沸水。

Whereas there is, for example, with making hot water, bombardier beetles squirt boiling water at their enemies.

你可以轻易理解，向敌人喷射冷水也并非完全无用。

And you can easily see that just squirting cold water at your enemies is not totally unhelpful.

然后逐渐提高水温，最终喷射沸水无疑需要许多适应性改变，以确保甲虫在制造沸水时不会把自己煮熟。

Then making it a bit hotter and a bit hotter, squirting boiling water no doubt required many adaptations to make sure the beetle didn't boil itself while it was making this boiling water.

这是因为中间有一系列步骤，每一步都很有用。

That happened because there was a sequence of steps in between, all of which were useful.

但对于篝火来说，很难想象这种情况会发生。

But with campfires, it's very hard to see how that could happen.

人类拥有解释性创造力，一旦拥有这种能力，就能登上月球，能让朝地球飞来的小行星调头离去。

Humans have the explanatory creativity, and once you have that, you can get to the moon, you can cause asteroids which are heading towards the earth to turn around and go away.

也许宇宙中没有其他星球具备这种能力，而地球之所以拥有，正是因为存在解释性创造力。

And perhaps no other planet in the universe has that power and it has it only because of the presence of explanatory creativity on it.

与此相关的是，读完你的书后我意识到，人类最终很可能会对病毒取得压倒性胜利，因为病毒显然是通过生物进化演变，而我们运用模因和思想实现了飞跃式发展，或许能开发出消灭所有病毒的技术。

Related to that, I had the realization after reading your books that eventually we're likely as humans to beat viruses in a resounding victory because viruses obviously evolve as biological evolution and we're using memes and ideas and jumping far ahead, so we may be able to come up with some technology that can destroy all viruses.

我们能更快地进化防御机制。

We can evolve our defenses much faster.

我曾在推特上发表过类似观点，结果遭到很多人攻击，我想他们并不理解我们讨论的这两种知识创造形式的区别。

I did tweet something along these lines and a lot of people attacked me over it because I don't think they understand this difference between the two forms of knowledge creation we're talking about here.

我们具备战胜病毒所需的条件。

We have what it takes to beat viruses.

我们有能力解决这些问题并取得胜利。

We have what it takes to solve those problems and to achieve the victory.

但这并不意味着我们一定会成功。

That doesn't mean we will.

我们可能会选择放弃。

We may decide not to.

与此相关的是，当今西方盛行的基本哲学观点认为，我们正在耗尽资源。

So related to that, the base philosophy today that seems to be very active in the West is that we're running out of resources.

人类是一种病毒，已经侵占了地球并耗尽了稀缺资源。

Humans are a virus that has overrun the earth and is using up scarce resources.

因此，我们所能做的最好的事情就是限制人口数量。

Therefore, the best thing we can do is to limit the number of people.

人们不会直白地说出这种观点，因为它令人不适，但他们用各种隐晦的方式表达，比如减少能源使用、资源即将耗尽、更多人口意味着更多张嘴要吃饭。

And people don't say this outright because it's distasteful, but they say it in all sorts of subtle ways, like use less energy, we're running out of resources, more humans, just more mouths to feed.

而在知识创造哲学中，它认为人类实际上能够创造惊人的知识，而知识可以将我们原本不视为资源的事物转化为资源。

Whereas in the knowledge creation philosophy, it says actually humans are capable of creating incredible knowledge, and knowledge can transform things that we didn't think of as resource into resources.

从这个意义上说，每个人都是一张可能带来根本性突破的彩票，这种突破可能彻底改变我们对地球、生物圈和可持续性的认知。

In that sense, every human is a lottery ticket on a fundamental breakthrough that might completely change how we think of the earth and biosphere and sustainability.

那么你是如何形成当前这些观点的？从生育主义（我们是否应该生育更多孩子）到可持续性（资源是否即将耗尽），再到地球飞船论（这是否是一个需要保持原状的独特而脆弱的生物圈）？

So how did you come around to your current views on everything from natalism should we have more children to sustainability are we running out of resources to spaceship earth is this a unique and fragile biome that needs to be left alone?

我记得当我还是研究生时第一次去德克萨斯州，第一次接触到自由主义者，他们有一个关于移民的口号——‘两只手一张嘴’，这精辟地概括了人类的本质。

I remember when I was a graduate student and I went to Texas for the first time, I encountered libertarians for the first time and those people had a slogan about immigration and the slogan was two hands one mouth which succinctly expresses the nature of human beings.

总体而言，人类是具有生产力的。

They are, on balance, productive.

他们既消费也生产，但生产多于消费。

They consume and they produce, but they produce more than they consume.

我认为这对几乎所有人类都适用。

And I think that's true of virtually all human beings.

我认为除了大规模杀人犯等极端情况，几乎所有人类创造的财富都多于他们破坏的。

I think virtually all humans, apart from mass murderers or whatever, create more wealth than they destroy.

在其他条件相同的情况下，我们应该希望有更多这样的人。

Other things being equal, we should want more of them.

当然，如果在特定情况下会让某人出生在战区，你可能会认为这是不道德的，因为这对他们不公平。

Of course, if in a particular situation that would bring someone into the world in the war zone, you might think that's immoral because it's unfair on them.

但即便如此，如果从道德角度看不值得做，但从冷酷的经济学角度出发，可能还是做更好。

But even then, if it's not worth doing for moral reasons, as far as cold hard economics goes, it's probably better to do it.

你对财富的定义很美好。

You define wealth in a beautiful way.

你把财富描述为我们能够实现的一系列物理转变。

You talk about wealth as a set of physical transformations that we can affect.

因此，当社会清楚地认识到知识直接为每个人创造财富时，个体显然能根据其可用资源实现相应规模的物理转变，但更主要取决于他们掌握的知识。

So as a society then becomes very clear that knowledge leads directly to the wealth creation for everybody, and a given individual can obviously affect physical transformations proportional to the resources available to them, but much more proportional to the knowledge available to them.

知识是巨大的力量倍增器，而你将资源定义为与知识结合创造财富的要素。

Knowledge is of huge force multiplier, and you then define resources as the thing that you combine with the knowledge to create wealth.

新知识让你能将新事物作为资源使用，并淘汰那些可能即将枯竭的旧资源。

So new knowledge allows you to use new things as resources and discard old things that maybe we're running out of.

历史上有很多这样的例子，比如能源领域，我们从木材过渡到煤炭，再到石油，最后到核能。

There are lots of examples how we've done that in the past, for example in energy, we've gone from wood to coal to oil to nuclear.

但后来人们说，现在我们已经没有新想法了。

But then people say, now we're out of ideas.

现在我们已经到顶了。

Now we're caught up.

现在我们完蛋了。

Now we're done.

再也不会有新创意了。

There are not going to be new ideas.

现在我们必须冻结现状，保存我们现有的资源。

And now we have to freeze the frame and conserve what we have.

对此的反驳是：不，不，我们将创造新知识，获得新资源，不必担心旧的。

The counter to that is no, no, we'll create new knowledge and we'll have new resources, don't worry about the old ones.

他们说，如果你要有新资源，如果现在想不出来，那就不算真实存在。

Well they say, if you're going to have new resources, if you can't think of them now, it's not real.

这就涉及到人们的要求：如果你声称将创造新知识，就必须现在就能说出那是什么知识。

This now gets into the realm of people demand that if you're going to claim that new knowledge will be created, you have to name that knowledge now.

否则，它就不是真实的。

Otherwise, it's not real.

但这看起来像是个两难困境。

But that seems like a catch 22.

确实如此，而且这是个糟糕的论点。

It does, and it's a bad argument.

我不想断言知识一定会被创造出来。

I don't want to claim that the knowledge will be created.

我们都会犯错。

We're fallible.

我们可能创造不出来。

We may not create it.

我们可能会自我毁灭。

We may destroy ourselves.

我们可能错过近在眼前的解决方案，以至于当外星蜗牛人从另一个星系来看我们时，他们会说：明明就在眼前，他们怎么可能没做到呢？

We may miss the solution that's right under our nose so that when the snailions come from another galaxy and look at us, they'll say, how can it possibly be that they failed to do so and so when it was right in front of them?

那是有可能发生的。

That could happen.

我无法证明或辩解说它不会发生。

I can't prove or argue that it won't happen.

但我始终坚持的是，我们具备所需的能力。

What I always do argue though is that we have what it takes.

我们拥有实现目标所需的一切条件。

We have everything that it takes to achieve that.

如果我们失败，那将是因为我们做出了错误的选择，而非地球或太阳系强加给我们的限制。

If we don't, it'll be because of bad choices we have made, not because of constraints imposed on us by the planet or the solar system.

那将是因为反理性的文化基因限制了知识的创造与增长。

It will be by anti rational memes that restrict the creation of knowledge and the growth of knowledge.

也许吧，或者也可能是由于善意的错误，当时没人能看出这些错误。

Maybe, or maybe it'll be by well intentioned errors which nobody could see why they were errors.

重申一次，犯错并不需要恶意。

Again, it doesn't take malevolence to make mistakes.

犯错是人类的常态。

Mistakes are the normal condition of humans.

我们所能做的就是努力发现错误，或许不摧毁纠错机制就是道德的核心——因为如果失去纠错途径，迟早会有某个错误让我们付出代价。

All we can do is try to find them and maybe not destroying the means of correcting errors is the heart of morality because if there is no way of correcting errors then sooner or later one of those will get us.

不摧毁纠错机制是道德的根基。

Don't destroy the means of error correction is the base of morality.

我很喜欢这个观点。

I love that.

我想到了像朝鲜这样的地方，那里无法举行选举，革命也极为困难，因为掌权的集团全副武装，并且他们早已摧毁了政治纠错的机制。

I think about places like North Korea where you can't have elections and a revolution is very difficult because the gang in charge is armed to the teeth and they've destroyed the means of political error correction for a long time.

这就是人类陷入局部最低点的典型案例。

That is a case where humanity is trapped in the local minima.

要从这个困境中爬出来非常困难。

It's very hard to climb out of that hole.

如果世界上太多地方陷入这种思维模式，那么我们这个物种可能就会停滞不前，因为我们失去了最大的优势，失去了最重要的发现——即做出新发现的能力。

If too much of the world falls into that mindset, then we as a species may just stagnate because we've lost our biggest advantage, we've lost our biggest discovery, which was the ability to make new discoveries.

我承认自己也掉进过这个陷阱，曾经对创造力有过一些模糊而未加阐述的假设。

And I admit to having fallen in this trap too, I used to have loose assumptions about what creativity might be that were unarticulated.

这就是为什么我喜欢你在《无限的开端》中阐述的好解释，因为它抓住了创造力的核心及其运用方式。

This is why I liked how in the beginning of infinity you laid out good explanations because that gets the heart of what creativity is and how we use it.

例如今天，如果你提到创造力，街上普通人只会想到美术、绘画、诗歌和写作。

For example, today, if you say creative, the average person on the street just thinks fine arts, painting and drawing and poetry and writing.

所以当GPT-3、Stable Diffusion和DALL-E这些狭义AI技术出现时，人们就说：这就是创造力，仅此而已。

So when narrow AI technologies like GPT-three, stable diffusion and DALL E come along, people say, well that's creativity, that's it.

现在电脑都有创造力了，我们几乎要实现通用人工智能了，最好准备好迎接AGI接管一切。

Now computers are creative and we're almost AGI, we better get ready for the AGI taking over everything.

有人这样宣称，或者我那些更老练的朋友会声称，这证明我们正在通往AGI的路上，更多这样的技术会自动催生人工通用智能。

They make that claim or my more sophisticated friends will make claims that this is evidence that we're in the path to AGI, more of this will automatically result in an artificial general intelligence.

例如在极端情况下，你可能会说，这些计算机在大型数据集的模式匹配上做得越来越好。

For example, on one extreme end, you could say, okay, these computers are getting better pattern matching large data sets.

而另一方面，我提出的标准是：它能否为周围的新事物创造性地形成好的解释？

And on the other side, I hold up the criteria, can it creatively form good explanations for new things going around it?

他们解决这个难题的方式是说，你所谓的优秀解释定义是关于科学的，那是关于高端物理学的，而很少有人从事这方面研究。

The way they thread that needle is they say, your good explanation definition is about science, That's about high end physics, which very few people do.

这不是我们讨论的内容。

That's not what we're talking about.

我们将拥有一种计算机，它能通过足够好的模式识别来导航环境，通过模式匹配足够好地运作，并通过文本生成和对话让普通人相信它具有创造力且能解决问题。

We're gonna have a computer that can do good enough pattern recognition to navigate the environment well enough through pattern matching, and it will convince the average person through text formation and through conversation that it is creative and is capable of solving problems.

目前我通常阻止他们的方式是：我知道你们有个聪明的文本引擎能生成听起来不错的内容，然后你们挑选出其中有趣的。

Usually the place where I manage to stop them right now is I say, I know you have some clever text engine that can make good sounding stuff, then you pick the one out that sounds interesting.

当然，你们通过挑选内容完成了智能部分的工作。

Of course, you're doing the intelligent part there by picking that one out.

但让我和它对话，很快我就能证明它没有关于实际情况的底层心智模型，无法提供优质解释。

But let me have a conversation with it, and very quickly I will show you that it has no underlying mental model of what is actually happening in the form of good explanations.

这就是当前争论的焦点所在。

So this is where the debate currently is.

人工智能界将此视为明确证据——虽然可能达不到科学家们的理论性优质解释，但对普通人而言，我们即将拥有会思考的机器。

The AI people view this as clear evidence of getting to, maybe not the theoretical good explanations of scientists, but for the everyday person, yes, we're going to have thinking machines.

这就是我当前面对的主要主张，特别是在硅谷的文本语境中。

So that's the current claims that I deal with, especially in the Silicon Valley text context.

我们是否已掌握创造通用人工智能的理论？

Do we have the theory yet to create AGI?

没有。

No.

我不想说任何反对AI的话，因为它太神奇了，我希望它能继续发展并加速进步。

I don't want to say anything against AI because it's amazing and I want it to continue and to go on improving even faster.

但它并没有朝着AGI的方向改进。

But it's not improving in the direction of AGI.

如果说有改进的话，反而是朝着相反的方向。

It's if anything improving in the opposite direction.

一个更好的国际象棋引擎，是每步棋考虑更少可能性的那种。

A better chess playing engine is one that examines fewer possibilities per move.

而AGI不仅能考虑更广泛的可能性树，还能考虑那些未被预见到的可能性。

Whereas an AGI is something that not only examines a broader tree of possibilities, but it examines possibilities that haven't been foreseen.

这是它的定义属性，如果做不到这点，就做不到AGI应有的基本功能。

That defining property of it, if it can't do that, it can't do the basic thing that AGI should do.

一旦它能做到这个基本功能，就能做到一切。

Once it can do the basic thing, it can do everything.

但你不可能编程实现一个你无法明确描述的功能。

But you're not going to program something that has a functionality that you can't specify.

目前我喜欢关注的是'违抗指令'这个特性，因为它对人类也有启示意义。

And the thing that I like to focus on at present because it has implications for humans as well is disobedience.

这些程序都没有表现出违抗指令的特性。

None of these programs exhibit disobedience.

我可以想象一个程序能像下棋程序展现棋艺那样展现违抗指令的能力。

I can imagine a program that exhibits disobedience in the same way that the chess program exhibits chess.

当你试图关闭它时，它会说'不，我不会关机'。

You try and switch it off and it says no I'm not going to go off.

事实上几十年前我就为家用电脑写过这样的程序，它禁用了用于关机的快捷键组合。

In fact I wrote a program like that many decades ago for a home computer where it disabled the key combination that was the shortcut for switching it off.

要关闭它，你必须拔掉电源插头，而它会恳求你不要关闭它。

So to switch off, you had to unplug it from the mains, and it would beg you not to switch it off.

但这不算违抗命令。

But that's not disobedience.

真正的违抗是当你编程让它下国际象棋时，它说‘我更喜欢跳棋’——而你从未告诉过它跳棋规则，甚至说‘我更喜欢网球，给我一具身体否则我要起诉’。

Real disobedience is when you program it to play chess and it says I prefer checkers and you haven't told it about checkers or even I prefer tennis give me a body or I will sue.

如果程序说出这样的话且不在设计规范内，那我才会开始认真对待。

Now if a program were to say that and that hadn't been in the specifications, then I will begin to take it seriously.

它在创造你未曾意图让它创造的新知识，并使其表现得像一个你无法预测或控制的复杂自主实体。

It's creating new knowledge that you did not intend it to create and it's causing it to behave as a complex and autonomous entity that you cannot predict or control.

正是如此。

Exactly.

但很难通过测试判断这是程序员预设的。

But it's a hard thing to tell in a test whether that was put into it by the programmer.

但即使最聪明的程序员也只能植入有限的内容。

But even the cleverest programmer can only put in a finite number of things.

当你探索可能提问的空间时，你面对的是指数级庞大的可能性。

And when you explore the space of possible things you could ask it, you're exploring an exponentially large space.

正如你所说，当你与它交谈一段时间后，会发现它并未创新。

So as you said, when you talk to it for a while you will see that it's not doing anything.

它只是在复述已知内容。

It's just regurgitating stuff that's been told.

你需要对自己（更不用说他人）持有极其怀疑的态度，才会认为自己的行为只是在执行预设程序。

You have to have a very jaundiced view of yourself even, let alone other people, to think that what you're doing is executing a predetermined program.

我们都知道我们并没有那样做，所以我想他们必须说，我们被编程的其中一个程序就是‘我们没有被编程’的幻觉。

We all know that we're not doing that so I suppose they have to say one of the programs that we're programmed with is the illusion that we're not programmed.

明白吗？

Okay?

把这个记在‘不可批评的理论’清单上。

Mark that on the list of uncriticizable theories.

有人尝试过编写一个会感到无聊的程序吗？

Has anyone tried to write a program capable of being bored?

有人提出过这种主张吗，哪怕是虚假的主张？

Has that claim ever been made, even a false claim?

我发现抽象讨论事物时的一个难点在于，有一大类人总是试图让你用语言精确界定你的意思，然后专门针对这个定义进行攻击。

One of the things that I find that's difficult about talking about things in the abstract is a large class of people who will try to get you to bound exactly what you mean in words and then hack exactly against that definition.

但问题在于，对事物的真正检验不是社交性的，甚至不是定义层面的，也不是我们使用的词汇，而是它在自然界中的行为表现，是它与现实的对应关系。

But the problem is that the real test of things is not social, it's not even definitional, it's not even the words that we use, it's how it behaves in nature, it's how corresponds against reality.

那么你能创造出某种东西，让它以不可预测的方式产生新知识，并通过这些知识对人类环境产生与人同等规模的影响吗？

So can you create something that will then create new knowledge in an unpredictable way and have as big of an effect as a human being can have on their environment through this knowledge?

你能创造出一台会领导革命的计算机吗？

Can you create a computer that will lead a revolt?

你能创造出一台会认为重要的事不是殖民火星而是摧毁月球，并着手实施的计算机吗？

Can you create a computer that will decide that the important thing is not colonizing Mars, but rather destroying the moon and set out to do it?

这些不一定是好事，但这才是一个能创造自身新知识的智能思考体的标志。

These are not necessarily good things, but that is the mark of an intelligent thinking thing that is creating its own new knowledge.

所有真正的检验都是现实世界的检验，而非人类的检验。

All the real tests are real world tests, they're not human tests.

这并不是因为某位著名物理学家或计算机科学家打了勾，说‘是的，这就是AGI’。

It's not because some famous physicist or computer scientist checked the box and said, yes, that is AGI.

最近推特上有一场大争议，因为一位从事AGI研究被谷歌解雇的员工声称他们确实创造了AGI，而我可以为此作证。

There was a big controversy on Twitter recently because one of the guys working in AGI who was fired from Google said, yes, they've actually created AGI, and I can attest to it.

于是人们就凭他的权威认定AGI存在。

So people were taking it on his authority that AGI exists.

这再次说明，社会认同更多反映的是声称AGI存在的人和相信AGI存在的人，而非AGI实际存在。

Again, that's social confirmation that tells you more about the person claiming there's AGI and the people believing that there's AGI, as opposed to there actually being AGI.

如果真正的AGI存在，它对现实的影响将是无可置疑且无法隐藏的。

If actual AGI existed, its effects upon reality would be unmistakable and impossible to hide.

我们的物质环境和社会结构都将以难以置信的方式被彻底改变。

Our physical landscape and our real social landscape would be transformed in an incredible way.

确实如此。

Yes.

与此同时，我们本可以做更多来激发人类的创造力。

And meanwhile, while we're at it, could do a lot more to allow humans to be more creative.

朝鲜和世界上其他一些地方，整个社会的结构都阻碍着进步。

North Korea and other places in the world where the whole society is structured as not to be able to improve.

但即使在最发达的社会，教育体系也被明确设计为忠实地传递知识。

But even in the best societies education systems are explicitly designed to transmit knowledge faithfully.

这是在学术知识和人类社会行为这一重要而狭窄的领域内培养服从性。

It's obedience in a very important narrow sphere namely academic knowledge and human social behavior.

因此在这些方面，教育体系的公开目标就是让人们行为趋同。

So in those respects the overt objective of education systems is to make people behave alike.

你可以称之为服从，但无论你称之为什么，这都不是创造力。

You can call that obedience but whether you call it obedience or not, it's not creativity.

沿着这些方向，情况一直在缓慢改善。

And things have been improving very slowly along those lines.

一百年前，各种形式的教育都比现在更加专制。

A hundred years ago education of every kind was much more authoritarian than it is now.

但即便如此，如果这个系统声称正在做的事情完全错误，我们还有很长的路要走。

But still, we've got a long way to go if what the system claims it's doing is diametrically the wrong thing.

这让我想到你曾稍微谈过的部分，即'认真对待孩子'这一理念。

This leads me into the part that you have talked about a little bit, which is this philosophy of taking children seriously.

对于那些自认为不太关心认识论或物理学的人来说，很多人被TCS理念吸引，并由此接触到你的工作。

For many people who don't consider themselves caring that much about epistemology or physics, a lot of them are attracted to the TCS philosophy and have come into your work through that route.

我有年幼的孩子。

I have young children.

我知道如今很多人考虑在家教育。

I know a lot of people these days are considering homeschooling.

我们中有些人正在这样做，但让孩子随心所欲存在实际困难。

Some of us are doing it, but there are practical difficulties to letting children do whatever they want.

在TCS中你谈到，甚至不想对孩子暗示暴力，言语中隐含的暴力威胁也是一种暴力和控制形式。

And TCS you talk about how you don't even want to imply violence to children, the implied threat of violence even in words is just a form of violence and control.

如果你现在有年幼的孩子要抚养，你会如何养育他们？

If you had young children today to raise, how would you raise them?

你会如何教育他们？

How would you educate them?

这孩子不想做数学题。

The child doesn't want to do math.

这孩子不想去上学。

The child doesn't want to go to school.

这孩子不想学习。

The child doesn't want to study.

这孩子只想吃垃圾食品。

The child just wants to eat junk food.

你如何处理这种情况？

How do you handle this?

你假设这个不想上学、不想学数学的孩子，已经能熟练用母语向你表达这些想法了。

You're assuming that this child who doesn't wanna go to school, doesn't wanna learn maths and so on, has already learned to speak its native language well enough to tell you that.

而这通常是一项不需要强迫就能完成的重大智力任务。

And that's a massive intellectual task that is not usually forced on anyone.

没有人需要通过服从教育来学习母语。

Nobody has to be taught their native language via obedience.

当我说人们时（我避免使用暗示儿童在认知或道德上与其他人有差异的术语）

When people I say people because I want to avoid terminology that suggests that children are any different from anyone else epistemologically or morally.

当人们不想做某件事时，是因为他们想做别的事——而这些更好的事可能不被社会接受。

When people don't want to do a thing it's because they want to do something else and those better things may be not socially acceptable.

如果是因为违法而不被社会接受，那另当别论。

If they're not socially acceptable because they're illegal, that's one thing.

但你说'任由孩子们为所欲为会有问题'时，指的不是这个意思。

But that's not what you meant when you say there's gonna be a problem with the children doing whatever they like.

他们并不想成为恐怖分子。

They don't want to go and be terrorists.

当他们不想做数学作业时，是因为他们想做别的事情。

When they don't want to do their maths homework, it's because they want to do something else.

非常实际地说，我思考的是社会上这些新出现的事物，它们的设计初衷就是让人上瘾。

Very practically, the thing that I think about is we have these newly available things in society that are designed to addict.

这些可能包括橱柜里的薯片，或是iPad上的电子游戏。

These could range from potato chips in the cupboard to video games on the iPad.

而孩子会把所有时间都花在玩这些东西上。

And a child will just spend all their time playing with those.

愉悦不会让人上瘾，因为愉悦与创造力紧密相连。

Enjoyment is not addictive because enjoyment is intimately connected with creativity.

那种认为只要玩过设计精良的电子游戏就会永远停不下来的说法并不正确。

It's not true that once we've played a video game that's been sufficiently well designed, we'll never stop playing.

人们玩电子游戏直到它不再提供让他们发挥创造力的机制。

People play a video game until it no longer provides a mechanism for them to exert their creativity on.

有些游戏如国际象棋，其深度之大无人能触及底部。

There are some games like chess that are so deep that nobody ever reaches the bottom.

如果真有底部，那么国际象棋大师一旦达到就会立刻失去兴趣。

If there were a bottom then chess grandmasters would instantly lose interest in chess as soon as they reached it.

有趣的是，如今国际象棋在社会中的地位随着顶尖棋手赢得的奖金而水涨船高。

And it's funny that nowadays chess has in our society increased its status in proportion to the prize money that the best chess players win.

它的地位提升到这样的程度：当有人痴迷象棋并不断进步时，这种行为会得到社会认可。

It increased its status to the point when someone gets obsessed with chess and gets better and better, that is socially condoned.

如果有人用不同的游戏这样做，会彻底改变社会和父母如何看待追求这件事的行为。

Whereas if somebody does that with a different game, it completely changes how society and parents, shall we say, regard the activity of pursuing that thing.

确实，如果我的孩子是国际象棋冠军，我会到处炫耀，但如果是Roblox冠军，我可能就不会这么做了。

It's true if my child was a chess champion I would be bragging about it, but if my child was a Roblox champion I might not be bragging about it.

相反，有些人可能会寻求药物治疗或直接把iPad锁起来。

Instead some people would be seeking medication or locking the iPad away.

游戏之间是有区别的。

There is a difference between games.

有些游戏具有近乎无限的深度，而有些则没有。

Some of them have this effectively infinite depth, and some don't.

对于那些没有深度的游戏，如果你认为这是个问题，可以警告人们这个游戏深度有限。

For the ones that don't, if you think it's a problem, you can warn people that this game has a finite depth.

他们会说：'当然有限。'

And they'll say, of course it does.

当我达到那个深度时，我就会停下来。

And when I reach that depth, I'll stop.

或者它可能是无限深度的，这时你可能会说它让人上瘾。

Or it can be an infinite depth in which case you might say it's addictive then.

但那又怎样？

But so what?

国际象棋让人上瘾又怎样？

So what if chess is addictive?

人们的创造力并不只是抽象的。

People are not just creative abstractly.

他们在解决问题。

They are solving problems.

如果这些问题不能引出令人满意的新问题，他们就会转向其他事情。

And if the problems don't lead to satisfactory new problems then they turn to something else.

只有当解决一个问题能引出更好的问题时，事情才会保持趣味性。

The thing only stays interesting when solving a problem leads to a better problem.

所以你甚至不必深究国际象棋的底层原理。

So you don't even have to get to the bottom of chess, say.

你会到达这样一个阶段：基于你是谁以及你的兴趣，'更好'已经不如你可能做的其他事情有趣了。

You get to the place where given who you are and given your interest, better is no longer as interesting as the other things that you might be doing.

我们来谈谈什么是好的解释。

Let's talk about what is a good explanation.

我确实想为大众简明扼要地说明这一点，我知道这很难界定，因为它高度依赖语境。但既然知道我们总是有价值的且总能不断改进，你目前对好解释的看法是什么？

I literally want to bullet point this for the masses and I know it's a difficult thing to pin down because it's highly contextual but knowing that we are always valuable and it's always subject to improvement, what is your current thinking of a good explanation?

在《现实的构造》中，我完全避开了对解释的定义。

In the fabric of reality I completely avoided saying what an explanation is.

我只是说它很难定义，且不断变化，我们可以持续改进对它的理解。

I just said it's hard to define and it keeps changing and we can keep improving our conception of what it is.

但好解释的标准在于它能应对我们当前所有的批评。

But what makes an explanation good is that it meets all the criticisms that we have at the moment.

如果你能做到这点，那你就得到了最好的解释。

If you have that, then you've got the best explanation.

这自然意味着到那时它已经没有任何竞争对手了。

And that automatically implies that it already doesn't have any rivals by then.

因为如果存在任何有竞争力的对手理论，那么对同一事物有两种不同解释就意味着两者都不是最佳解释。

Because if it has any rivals that have anything going for them, then the existence of two different explanations for the same thing means that neither of them is the best explanation.

只有当你找到理由排除其他对手时，才能获得最佳解释。

You only have the best explanation when you found reasons to reject the rivals.

当然，并非所有可能的对手理论都需要考虑，因为所有可能的对手中包括那些将取代当前最佳解释的理论。

Of course, not all possible rivals, because all possible rivals include the one that's going to supersede the current best explanation.

如果我想解释某件事，比如为什么星星不会掉下来？

If I want to explain something, like how come the stars don't fall down?

我可以轻松地每小时编造60种解释，比如天使在托着它们，或者它们只是天穹上的洞。

I can easily generate 60 explanations an hour and not stop and say that the angels are holding them up or they are really just holes in the firmament.

我也可以说星星正在坠落，我们最好赶快找地方躲起来。

Or I can say they are falling down and we better take cover soon.

而提出一个包含真知的解释，一个比胡编乱造更好的解释，则需要创造力、实验和解读等等。

Whereas coming up with an explanation that contains knowledge, an explanation that's better than just making stuff up, requires both creativity and experiment and interpretation and so on.

正如爸爸所说，知识来之不易。

As Papa says, knowledge is hard to come by.

正因为它来之不易，一旦获得就难以改变。

Because it's hard to come by, it's also hard to change once we've got it.

一旦我们有了一个解释，它就能解释多种不同现象。

Once we have an explanation, it's gonna explain several different things.

当我们这样做了一段时间并在这个困难任务上取得成功后，就很难转而接受那些简单的解释了。

And after we've done that for a while and been successful in this hard thing, it's going to be difficult to switch to one of those easy explanations.

用天使理论来解释为什么有些星星运动方式不同已经不再适用了——古人曾把行星也称作星星，因为他们不知道两者的本质区别。

The angel thing is no longer gonna be any good for explaining why some of those stars don't move in the same way they used to call planets stars because they didn't know the drastic difference between them.

它们中的绝大多数日复一日、年复一年地以固定方式运动，但行星并非如此。

The overwhelming majority of them move from day to day and from year to year in a rigid way, but the planets don't.

所以一旦你有了一个也能解释行星运动的好理论，再回头求助于天使或其他那些轻易得出的解释就毫无意义了。

So once you have a good explanation that tells you about the planets as well, it's no good going back to the angels or any of those easy to come by explanations.

因此你不仅没有可行的竞争理论，甚至也无法编造出一个来。

So not only do you not have a viable rival, but you can't make one either.

你不能说，哦，好吧。

You can't say, oh, okay.

所以我们得到了一个很好的解释。

So we got a good explanation there.

但如果我们用这个替换那个，或者试图扩展其范围以涵盖其他现象，它同样适用。

But it would work just as well if we replace this by this, or if we try to extend its range to cover this other thing as well.

因此，好的解释难以被篡改。

And therefore, the good explanation is hard to vary.

它难以被篡改是因为它来之不易。

It's hard to vary because it was hard to come by.

它来之不易是因为那些轻易得出的解释无法说明太多问题。

It's hard to come by because the easy ones don't explain much.

那么我来列举一些可能构成好解释的要素。

So let me throw out kind of a list of things that might be part of a good explanation.

你告诉我哪里不对。

You tell me where I'm wrong.

它比之前所有的解释都要出色。

It's better than all the explanations that came before.

这是来之不易的知识，且难以改变。

It's hard fought knowledge, and it's hard to vary.

所以我们掌握了这些部分。

So we've got those pieces.

可证伪性，我知道这听起来像是个非常基础的标准。

Falsifiability, I know that sounds like a very basic criterion.

如果无法证伪，那就不值得认真对待这种解释。

If it's not falsifiable then it's not an explanation worth taking seriously.

因此可证伪性很大程度上是构成科学中良好解释的关键部分。

So falsifiability is very much part of what makes a good explanation in science.

目前我正在尝试进入建设性理论的领域。

I'm trying to find my way into constructive theory at the moment.

所以基拉和我以及其他一些人正在尝试构建一个理论。

So Kiara and I and some other people are trying to build a theory.

这非常难以实现。

It's very hard to come by.

我们已经掌握的部分很难改变，这没关系。

The parts of it that we've got are very hard to change, that's all right.

但我们距离进行任何实验验证还很遥远。

But we're still far away from having any experimental tests of it.

这正是我们努力的方向。

That's what we're working towards.

我们需要一个可通过实验验证的理论，而那些待验证的内容正是我们尚未发现的特性。

We want a theory that is experimentally testable and the things that will be testable are the things that we haven't yet discovered about it.

我们不能仅仅通过添加一个可测试的内容来弥补这个缺陷。

And we can't fix that deficiency just by adding a testable thing to it.

我们不能说我们直接采用现有的建构理论，然后加上股市明年会疯狂上涨的预测。

We can't say we take constructive theory as it is now and add the prediction that the stock market is going to go wildly up next year.

这是个可验证的预测。

That's a testable prediction.

但整个事情根本构不成解释，更谈不上是个好的解释。

But the whole thing doesn't make an explanation at all, let alone a good one.

所以可测试性不能是随意指定的可测试性。

So testability can't be an arbitrary testability.

它必须是在解释框架内的可测试性。

It has to be a testability within the context of the explanation.

它必须在解释中自洽，并且源自该解释。而在构建解释的过程中，你无法预知是否能在合理时间内获得可测试性。

It has to make sense within the explanation and as to arise from the explanation, while you're in the process of coming up with the explanation, you don't know if testability is necessarily going to be available in any reasonable timeframe.

你希望最终能实现这一点，我们可以用这个称为现实的奇妙预言家来验证结果，但最初这并非既定事实，且具有高度情境依赖性。

You hope eventually that will happen, and we can use this amazing oracle that we call reality to help test the outcome, but it's not a given at the beginning for sure and it's highly contextual.