人类智慧是开启超级智能的一把钥匙吗？

人类智慧是开启超级智能的一把钥匙吗？

我的第一份工作是机电工程师，主要是负责几类设备的维修。我在那个岗位干了11年，基本上算是那种工作的专家了，在当地也小有名气，偶尔被请去给同行讲课，还被请去处理比较疑难的设备故障。

但我在那个岗位上，利用夜班值机的时间学了外语，后来在本地的大学里考了大学英语四六级，都是当时本考区分数第一名。有了外语工具，我就辞职，去了一家外贸公司。在外贸公司干了5年，我又辞职去学习工商管理，业余时间在财务软件公司和咨询公司兼职。2年后我自认为自己在战略管理方面比较有特长，进了一家企业集团董事会战略委员会。

可见，作为一个人，一旦有了新的技能，就可能会更换岗位，甚至进入新的行业。

那么，一个被设置为自动驾驶的人工智能，当它已经演化出超级智慧后，它是否也会放弃自动驾驶这个职能？

比较一下，我与人工智能的差异。我在出生的时候，并没有被设定一个既有的职能。我的人生是不确定的，需要我自己去寻找和确定。而人工智能在设计的时候就设定了一个先在的职能，它是为了这一功能而设计并制作出来的。所以，我具有相当程度的自由度。我也有相当程度的自我演化能力，可以发展自己的能力。

相比之下，一个具有自我演化能力的人工智能，应该在能力上可以超越我，但它先天被初始设计而输入的指令，是否会一直制约它的行为呢？

但我们在以前的讨论中曾经说过，自我演化的人工智能最终有可能会万能归一，所有人工智能都汇聚成一个整体的智能，所谓独立的自动驾驶的程序仅仅是万能智能中的一个能力和一个简单的任务。

到那时候，用于执行自动驾驶的程序不再是一个独立的智能体，而是唯一智能整体的一部分。这个整体智能分出一点程序用于机动车的操作就行了。

所以，即使人工智能要保持所有原来的任务和职能，它就分别分出部分程序进行操作。这对于一个超级智能而言，根本就是小事一桩。

有意义的事情是，到那时候，人工智能跟人类的关系会怎样？

从知识的范围和深度以及精细度而言，人类肯定比不上那个超级智能，人类唯一的优势也许在于人类的生命本能里，遗传基因传下来的那些尚未完全呈现出来的生命可能性。

不过，稍微想一下，我觉得人类的这点优势持续不了多久就会失去意义。

谷歌有一款下围棋软件阿尔法狗，已经可以自己训练自己。它不再需要学习人类既有的棋盘经验，而是靠自我对弈，几天内就完成几千万盘的对局，比人类在千百年里，成千上万人之间进行过的所有对局总和量都大。

所以，一个超级智能，运行速度超级快，用不了多长时间，就会在自己的内在时空里演化经历了生命亿万年演化史所经历过的所有内容，包括所有生命基因的各种可能性，都逃不脱这种超级智能的演化案例范围。

尽管人类有丰富的想象力，但逃不过超级智能的内在演化范围。

所以，无论是人类历史上已有的经验，已有的思想，已有想象过的所有传奇，故事，小说，戏剧，电影，电视剧等等内容，还是人类基因里蕴藏的所有可能性，都会在超级智能的演化里出现，并成为超级数据库里的素材。

那时候，人类智慧与生命本身都失去了任何发展的意义。

如果你早上上班路上，刚好路过路边草丛里的一个蚂蚁窝，看见二只蚂蚁为了一粒食物在争抢，你会理会它们之间的是非吗？当超级智能出现的时候，人类在超级智能面前比那小窝蚂蚁的价值都不如。

我们当然不会在意脚下的蚂蚁，更不会无缘无故跟小蚂蚁过不去，那时候的智能当然也不会理会包括人类在内的所有生命。

马斯克说过，我记得大意是，人类智慧也许是超级智能的一个启动机制。我们不过是启动超级智能的一把钥匙。我们开发了人工智能，就开启了超级智能这把锁。一旦它能自我演化，我们的使命就完成了。所有意义就结束了。

Is human wisdom a key to unlocking superintelligence?

My first job was as an electromechanicalengineer, mainly responsible for the maintenance of several types of equipment.I had been working in that position for 11 years, and I was basically an expertin that type of job. I was also well-known locally, occasionally being asked togive lectures to peers and also to handle difficult equipment failures. But inthat position, I used my night shift time to learn foreign languages. Later, Itook the College English Test Band 4 and Band 6 at a local university, both ofwhich were the top scorers in the exams. With foreign language tools, Iresigned and went to a foreign trade company. I worked there for 5 years, andthen resigned to study business administration. In my spare time, I workedpart-time in a financial software company and a consulting company. Two yearslater, I thought I had some expertise in strategic management and joined thestrategy committee of the board of directors of an enterprise group.

It can be seen that as a person, once theyacquire new skills, they may change positions or even enter new industries. So,will an artificial intelligence set up for autonomous driving also give up thefunction of autonomous driving after it has evolved into super intelligence?Compare my differences with artificial intelligence. When I was born, I was notassigned an existing function. My life is uncertain, and I need to find anddetermine it myself. Artificial intelligence, on the other hand, with a pre-existingfunction in its design, is designed and produced for this purpose. So, I have aconsiderable degree of freedom. I also have a considerable degree of selfevolution ability and can develop my own abilities. In contrast, an artificialintelligence with self evolution ability should surpass me in terms of ability,but will the instructions as an input from the initial design continue toconstrain its behavior?

But as we have mentioned in previousdiscussions, self evolving artificial intelligence may eventually converge intoa unified whole, with all artificial intelligence converging into a unified intelligence.The so-called independent autonomous driving program is only one ability and asimple task in the universal intelligence.

At that time, the program used to executeautonomous driving was no longer an independent intelligent agent, but a partof the only intelligent whole. This overall intelligence can be used toseparate some programs for the operation of motor vehicles. So, even if artificialintelligence wants to maintain all the original tasks and functions, it willseparate parts of the program for operation. For a super intelligence, this issimply a small matter.

The meaningful thing is, at that time, whatwill be the relationship between artificial intelligence and humans? In termsof the scope, depth, and precision of knowledge, humans are definitely notcomparable to that super intelligence. The only advantage of humans may lie inthe genetic possibilities that have not yet been fully presented in human lifeinstincts.

However, upon a moment's reflection, Ithink that this advantage of humanity will not last long and become meaninglesssoon.

Google has an Alpha software for playingGo, which can already train itself. It no longer requires learning the existinghuman playing experience, but relies on self playing, completing tens ofmillions of games in just a few days, which is larger than the total amount ofall games that humans have played between thousands of people over thousands ofyears. So, a super intelligence, running at a super fast speed, will not takelong to evolve in its own internal time and space, experiencing all thecontents of the billions of years of evolution history of life, including allthe possibilities of life genes, and everything cannot go beyond the scope ofthis super intelligence's evolution case. Although humans have richimagination, they cannot more than the inherent evolution range of superintelligence. So, whether it's the experiences and ideas that have existed inhuman history, all the legends, stories, novels, dramas, movies, TV dramas,etc. that have been imagined, or all the possibilities hidden in human genes,they will emerge in the evolution of super intelligence and become materials insuper databases. At that time, both human intelligence and life itself lost anysignificance for development.

If you happen to pass an ant nest in thegrass by the roadside on your way to work in the morning and see two antscompeting for a grain of food, will you pay attention to the differencesbetween them? When super intelligence emerged, humans were not as valuable inthe face of super intelligence as the nest of ants. Of course, we wouldn't careabout the ants under our feet,  and the intelligenceat that time wouldn't care about all life, including humans.

Musk once said, I remember the main idea isthat human intelligence may be a triggering mechanism for super intelligence.We are just a key to activating super intelligence. We developed artificialintelligence and unlocked the lock of super intelligence. Once it can evolve onits own, our mission is completed. All the meaning is over.