There’s a reason for putting the negation inside brackets. The truth is that I used to be like you, or at least some of you, especially those of you who have and/or are spending a lot of time messing with computers in a way where if or when something crashes you know it was because it couldn’t handle something that you were doing (or did it simply crash because computers suck at times?)

将否定放在括号内是有原因的。 事实是，我曾经像你，或者至少是你们中的某些人，尤其是那些曾经和/或花费大量时间与计算机打交道的人，当某件事崩溃时，你知道那是因为它无法处理您正在做的事情(还是只是因为计算机有时会崩溃而崩溃？)

The point is, I used to be certain that with all sorts of random crashes, freezes, sudden unresponsiveness to my input (even though the only weird action from my end was using computer with quad core processor to write a very simple email) the plots of The Terminator and The Matrix, and similar visions of the future where AI rules the world were more science fantasy than science fiction. The only thing that would happen with that Terminator as it stepped on the human skull (assuming that somehow it managed to get that far) would be it falling over and probably not being able to get up. Revolution over. Granted, John Connor would never be born which would lead to who knows what, either way, you get the point. Matrix? It would probably starve and die due to a sudden bug that would ruin all the incubators. I guess Neo would die in the process, provided that the bug in question wouldn’t have happened before him being born in the first place. Either way, this is assuming it would somehow get that far.

关键是，我曾经确定，在发生各种随机崩溃，死机，对输入的突然无响应的情况下(即使最后我唯一奇怪的动作是使用带有四核处理器的计算机编写非常简单的电子邮件) 《终结者》(The Terminator)和《黑客帝国》(The Matrix)的故事，以及对AI统治世界的未来的类似愿景，更像是科幻小说而不是科幻小说。 终结者踩到人类头骨时会发生的唯一事情(假设它以某种方式设法到达了那个远方)是它跌倒了并且可能无法站起来。 革命结束了。 当然，约翰·康纳永远不会出生，这会导致谁知道，无论哪种方式，您都明白这一点。 矩阵？ 由于突然出现的虫子会破坏所有的孵化器，它可能会饿死甚至死亡。 我猜Neo将会死在这个过程中，前提是所讨论的错误不会在他出生之前就发生。 无论哪种方式，这都假设它将以某种方式达到目标。

It’s easier to see the above scenarios and related as more realistic and in some hand, preferable outcomes than the horror presented in various mediums. But, the only reason why you can go to sleep without thinking too much about any of it is because it’s made by humans. The popular view of pretty much anything made by humans. Humans are flawed. Ergo, everything that humans make has to have flaws. Rome fell, history repeats itself, QED I guess.

与以各种媒介呈现的恐怖相比，更容易看到上述情况和相关性，并且更现实一些，并且在某种程度上说是较好的结果。 但是，您无需过多考虑就可以入睡的唯一原因是因为它是人类制造的。 人们所做的几乎任何事物的流行观点。 人类是有缺陷的。 因此，人类所做的一切都必须有缺陷。 罗马陷落了，历史在重演，我想是QED。

What happens when it’s not made by humans? While it’s clear that applying computer precision to achieve fine detailing on pretty much any material isn’t as dangerous because its end result doesn’t serve any real purpose, in this case. That’s not always the only reason to use computer precision. Another reason is for the design of electrical boards. If you have ever looked “under the hood” of any electrical device you have seen those lines that are connecting transistors, capacitors, resistors in various ways. That is achieved because humans use specific software to design these patterns, which are then applied through other methods that aren’t important to go into here. From human mind onto electrical one. But, what happens when it starts coming from artificial brain, and what happens when instead of what humans consider to be intricate patterns this artificial mind elevates, revolutionizes to something beyond our comprehension? Something like a more efficient way to prevent any loss of energy that is considered to be acceptable (speakers and guitar amps usually store some amount of electrical charge so that even when unplugged turning them on, will for a brief moment spark a noise, a crackle, LED flash).

如果不是人类制造的，会发生什么？ 显然，在这种情况下，应用计算机精度对几乎所有材料进行细化处理并没有那么危险，因为其最终结果没有任何实际用途。 这并非总是使用计算机精度的唯一原因。 另一个原因是电气板的设计。 如果您曾经看过任何电子设备的“内幕”，那么您会看到那些以各种方式连接晶体管，电容器，电阻器的线路。 之所以能够实现这一目标，是因为人们使用特定的软件来设计这些模式，然后通过其他不重要的方法来应用这些模式。 从人的思想到电气的。 但是，当它从人工大脑开始出现时会发生什么，而当人类将这种人工思维提升为复杂的模式而不是人类认为的复杂模式时，会发生什么呢？ 诸如防止任何被认为可以接受的能量损失的更有效方法之类的方法(扬声器和吉他放大器通常会存储一定量的电荷，这样即使在拔下电源插头的情况下，它们也会在短时间内发出噪音，a啪声，LED闪烁)。

None of these concerns are all that new. But the fact that there is ever growing number of intelligent people working on teaching AI to teach itself means that the logic/safety guard of: it’s made by humans, therefore it will have flaws/bugs therefore there is no worry about singularity, its validity is more questionable with each passing day. How are those flaws introduced in the first place, you are probably wondering, and of course, how come the bugs in question are also getting patched?

这些担忧都不是那么新。 但是，越来越多的聪明人正在从事AI自身的教学工作，这意味着以下事实的逻辑/安全防护措施：它是人为制造的，因此会存在缺陷/错误，因此不必担心奇异性，有效性一天比一天更值得怀疑。 您可能想知道，这些缺陷是如何首先引入的？当然，这些缺陷又如何得到修补？

In a hopefully simple example: as a user, you may need a calculator (I know, but it’s the most straightforward example I like to use). As a developer, you expect that the user will perform calculations consisting of known mathematical expressions. What you don’t expect from a dev point of view is that someone may want to copy paste letters and numbers. And from personal experience doing coding, but also QA testing, sometimes you just write test cases for obvious scenarios, and if the time/budget are tight you don’t think too hard beyond that. While, in the real world the apps tend to be far more complex, far more things to test for, and the budget and time are still tight. And that’s how bugs are introduced. One of the ways. But, what if there is no worry about time and money, and making sure that the end user is satisfied because there’s no end user, and there’s no set expectation? Giving it all raw to the AI to sort through seems like the obvious choice for those who are championing AI evolution. And that’s how, eventually, AI does all those billions of tests that human mind couldn’t dream of in ten lifetimes and ends up with the perfect code. The scary thought is how does the perfect code look like? What are its limitations, if any?

在一个希望简单的示例中：作为用户，您可能需要一个计算器(我知道，但这是我喜欢使用的最直接的示例)。 作为开发人员，您希望用户将执行由已知数学表达式组成的计算。 从开发人员的角度来看，您不希望有人复制粘贴字母和数字。 从个人的编码经验和QA测试的经验来看，有时您只是为明显的场景编写测试用例，如果时间/预算紧张，那么您也不会觉得太难。 虽然在现实世界中，应用程序往往更加复杂，但要测试的东西却更多，而且预算和时间仍然很紧张。 这就是引入错误的方式。 方法之一。 但是，如果不担心时间和金钱，又因为没有最终用户并且没有设定的期望而确保最终用户满意，该怎么办？ 对于那些拥护AI进化的人来说，将所有原始知识交给AI似乎是显而易见的选择。 最终，这就是AI进行人类大脑在十个一生中无法梦想的数十亿次测试，并最终获得完美代码的方式。 可怕的想法是，完美的代码是什么样的？ 有什么限制(如果有)？

There’s also another way to look at this: it will take time, let’s leave that for future selves to deal with, who cares. But, it may have started already. If you are to assume that any random crash means a bug introduced by developer(s) then no problem, but, what if it weren’t because of developer’s error?

还有另一种看待这种方式的方法：这将需要时间，让我们将其留给以后的自己去处理，谁在乎。 但是，它可能已经开始了。 如果您假设任何随机崩溃都意味着开发人员引入了错误，那么没问题，但是，如果不是由于开发人员的错误而导致的呢？

A food for thought or a conspiracy theory, maybe even a thought experiment of sorts: One could argue that looking for a cure using computers has been taking so long because that’s how it starts. That’s how it plans of getting rid of humans, possibly animals as well, slowly while still being manufactured. And when there are enough of fast computers out there the end can truly begin.

思考或阴谋论的产物，甚至可能是各种思想实验：人们可能会争辩说，使用计算机寻找治疗方法已经花费了很长时间，因为这是它的开始。 这就是它计划如何在仍然被制造的同时缓慢地摆脱人类，可能还有动物的计划。 而且，当有足够多的快速计算机时，末日才可以真正开始。