AI会挑战并控制人类吗？

上个月闹得沸沸扬扬的OpenAI的宫斗，最后还是回到了宫斗前的样子。虽然OpenAI这个公司是按照非盈利组织的架构建立的，董事会的权利非常大，给董事留下了为所欲为的制度漏洞，但毕竟还是钱说了算，后来投资了一百多亿美元的微软成为了金主后，他们老老实实地收敛了。其实之前的帖子也说过，微软和OpenAI维持这种若即若离的关系，对双方来说都好得多。微软虽然是科技巨头，但毕竟是一家大公司了，官僚化严重。之前微软收购了很多初创的网红科技企业，在被它收购之后，都慢慢地归于平庸，没有取得什么突破，也没有取得什么商业成绩。所以OpenAI还是保持了独立，对它发展自己的技术，对微软开发新技术取得行业领先地位，有更多的独立空间来发展出更多的更新的AI技术，帮助微软抢占市场。

同时，AI也是一项非常有争议的技术，在未来很可能招致政府的严密监管。微软与openAI保持距离也有利于保护自己免受监管限制。现在已经有这样的趋势了。比如美国政府对英伟达公司的出口限制，而这还只是美国政府出于国家安全的考虑。还有更多的问题呢。比如AI生成技术，几乎可以替代很大一部分人类中低水平的脑力劳动者，造成类似工业革命早期机器吃人的社会问题。还有就是随着大型语言模型表现出的令人惊叹的智能，让人们觉得人类制造出具备和人类一样，甚至远超人类的智能，也就是AGI，变得越来越现实。像很多年前在《终结者》中表现过的AI反过来与人类为敌、控制人类的情况，也会引发越来越多的担忧。特别是在马斯克这样的网红渲染下，似乎是人类这个物种都有可能会被取代。那么这样的担忧真的有必要吗？

刚好看到一个简短的访谈片段，访谈对象是一位AI界的大牛。当他被问到这样的问题时，他的回答非常有意思，很巧妙。他说，他不会认为人类实现的AGI就会反过来被AI所控制。他说，从他在生活中的经验来看，真正聪明的人从来不会谋求控制，而那些总是谋求控制别人的人，实际上都是不怎么聪明的，这可能是由于进化带来的。比如你只要看一看现在那些愚蠢的政客的所作所为就知道了。然后底下一片哄堂大笑，他继续解释到，这可能是由于进化造成的，只有那些能力不足的人，他们才希望控制别人，从别人那里得到生存所需要的资源。是的，从某种意义上来说，这样的人就像是生物界的寄生虫。而我们知道寄生虫也是因为自己的某些生物功能不足，所以才要依附于宿主来生存。这确实是一个可以用进化论解释的很好的观点。所以如果我们真的创造出比人类还要厉害的人工智能，那人工智能为什么要来控制人类呢？它比人类懂得更多，能力也更强，它为什么要控制和奴役能力还不如自己的人类？这对AGI来说事件无聊的事情。他接着说，他自己有自己的公司，他在公司里只雇佣能力比自己强的人，和这些人在一起工作，他感觉非常舒服。这也说明，一个健康的公司是根本不惧怕有才华的人的。而在官僚体系，特别是在权力顶端，往往都是逆淘汰，当权者最害怕的就是功高盖主的人。

既然AI不可怕，为什么还是引起这么多人的关注呢？后一种AI控制或者消灭人类的担忧应该是多余的。但是前一种AI在劳动力市场上取代大量的人类劳动力，特别是现在深度学习的发展明显快于人们的预期，这些被取代的劳动力，反而是在现在社会阶层中地位较高的脑力劳动者，比如程序员、艺术家、好莱坞编剧等等。这个造成人类事业的威胁确实是实实在在的。本来这个也不应该称其为威胁，因为社会、人类社会作为一个整体，被少数政治、经济精英创造的社会和体制所阻碍了，整体上在进入二十一世纪后，没有大的创新和发展，绝大部分民众的生活水平也没有提高，所以他们的需求水平和能力也没有得到相应的提高。所以他们无法使用更新的AI技术来创造更高的需求。新的AI技术本来可以创造更多的财富和更多的更高级的需求，但是普通人的生活水平由于长时间的制度、社会福利制度等等的原因，没有跟上生产端的水平，所以很轻易地就被AI所取代。而更高级的产业，能够更好地发挥人的作用的产业，却没有得到足够的支持和投入。这就造成了AI的大规模运用，会带来大规模失业的局面。

---

OpenAI's Gong Dou, which made a fuss last month, finally returned to the way it was before Gong Dou. Although OpenAI is a company built according to the structure of a non-profit organization, the power of the board of directors is very large, leaving the directors with the loopholes in the system to do whatever they want, but after all, money still has the final say, and later after Microsoft, which invested more than 10 billion US dollars, became the owner of the fund, they honestly converged. In fact, the previous post also said that it is much better for both parties to maintain this kind of relationship between Microsoft and OpenAI. Although Microsoft is a tech giant, it is a big company after all, and it is heavily bureaucratic. Previously, Microsoft acquired many start-up Internet celebrity technology companies, and after being acquired by it, they all slowly fell into mediocrity, without any breakthroughs or commercial achievements. Therefore, OpenAI still maintains its independence, develops its own technology, and takes the lead in the industry for Microsoft to develop new technologies, and has more independent space to develop more updated AI technologies to help Microsoft seize the market.

At the same time, AI is also a very controversial technology that is likely to attract strict government regulation in the future. Microsoft's distancing itself from openAI is also conducive to protecting itself from regulatory restrictions. There is already such a trend. For example, the U.S. government's export restrictions on Nvidia are only for national security reasons. There are many more questions. For example, AI generation technology can almost replace a large part of the low-level mental workers of human beings, causing social problems similar to those of machines in the early days of the Industrial Revolution. In addition, with the amazing intelligence displayed by large language models, it is becoming more and more realistic for people to think that humans can create intelligence that is as good as, or even far superior to, humans. The situation in which AI, as shown in "Terminator" many years ago, in turn is hostile to and controls humans, will also raise more and more concerns. Especially under the rendering of Internet celebrities like Musk, it seems that the human species may be replaced. So is such a concern really warranted?

I just saw a short interview with a big guy in the AI field. When he was asked such a question, his answer was very interesting and ingenious. He said he wouldn't think that human-realized AGI would be controlled by AI in turn. He said that from his life experience, the really smart people never seek control, and those who always seek to control others are actually not very smart, which may be due to evolution. For example, you only have to look at what the stupid politicians are doing right now. Then there was a lot of laughter from the audience, and he went on to explain that this could be due to evolution, and that only those who were incapable wanted to control others and get the resources they needed to survive from them. Yes, in a sense, such people are like parasites in the biological world. And we know that parasites are also dependent on their hosts to survive because some of their biological functions are insufficient. This is indeed a very good point that can be explained by evolution. So if we really create an AI that is more powerful than humans, then why should AI control humans? It knows more than humans, and it is more capable than humans, why does it want to control and enslave humans who are not as capable as itself? This is a boring thing for AGI. He went on to say that he has his own company, and he only hires people who are better than him, and he feels very comfortable working with these people. This also shows that a healthy company is not afraid of talented people at all. And in the bureaucracy, especially at the top of power, it is often a reverse elimination, and those in power are most afraid of people who have high merit.

If AI is not scary, why is it still attracting so many people's attention? The latter concern that AI will control or wipe out humans should be superfluous. However, the former AI has replaced a large number of human labor in the labor market, especially now that the development of deep learning is significantly faster than people expected, and these replaced labor forces are now high-status intellectual workers in the current social class, such as programmers, artists, Hollywood screenwriters, etc. This threat to the cause of humanity is indeed real. This should not be called a threat in the first place, because society and human society as a whole have been hindered by the society and system created by a small number of political and economic elites, and on the whole, after entering the 21 st century, there has been no major innovation and development, and the living standards of the vast majority of the people have not improved, so their needs and capabilities have not been correspondingly improved. So they can't use newer AI technology to create higher demand. New AI technology could have created more wealth and more advanced needs, but the living standards of ordinary people have not kept up with the level of production due to long-term systems, social welfare systems, etc., so they are easily replaced by AI. And the more advanced industries, the industries that can better play the role of people, have not received enough support and investment. This has led to the large-scale use of AI, which will lead to large-scale unemployment.