人类智能比人工智能更危险？

最近，我读了一本名为《千脑智能》的书。作者的主要观点是，人工智能并不可怕，人类智能才是最可怕的。虽然书中也有一些新颖和有趣的观点，但是感觉作者还是夹带了很多私货。从他的举例和论证来看，他是一个典型的白左。他坚信气候变化是人类活动造成的，并且将会毁灭整个地球，所以他经常拿这个事实来指责异见者。让我们来看看他是如何论述人类智能比人工智能更危险的吧。
作者的智能理论将人类的大脑分为旧脑和新脑。旧脑是由漫长的进化所形成的，这一过程长达数十亿年。这一过程完全是由基因主导的。作者很明显受到了理查德·道金斯的《自私的基因》这本著作的影响。他认为生物不过是基因的生存机器，生物的基本功能就是生存下去，并且在繁殖过程中让基因得以延续。而旧脑的功能就是围绕着这一目的进化的。因为基因本质上就是一串只知道复制的DNA大分子，它没有情感，也没有道德感，也不会感受痛苦。所以它不会区分行为的生存策略的善恶，只要能够保证它的复制和传递就是好的策略。于是旧脑就形成了让我们产生各种情感，会照顾幼子，不同个体之间能够合作分工等等行为。这些行为对基因的存续是有帮助的，但是同时抢夺资源，暴力攻击，强奸等行为也能够达到让基因存续的目的。而我们旧脑是无法区分这些手段的善恶的，只有新脑的新皮质掌握了复杂的世界模型，能够做出更精确和长远的推测。才会知道某些行为虽然短期能够带来收益，长期却会破坏人类的生存环境和社会稳定。
但是旧脑因为是早期进化的结果，并且从底层上控制着身体的行为，所以人类很难克制它。就像很多人知道，食用太多的甜食和碳水化合物会损害健康，增大患心脑血管疾病的风险，但当一块美味的蛋糕放在你面前的时候，多半还是会把它吃下肚里。

另一个典型的例子就是人口控制，这也表现出了作者的白左观点。他认为人口的无限增长会最终摧毁整个地球，应该实行计划生育。因为生育的本能是由基因驱动的，也是很难克服的。不过作者说，现在的人类发明了很多的避孕措施，他将这解释为新脑在利用智能对抗旧脑的本能。不过，作者的观点虽然偏向白左，但是有一点还是不错的。对于人口问题，他提出了一项自认为可行的解决方案，而且并不像我国曾经实行过的计划生育那样完全依靠政府的强制力来推行。他说这个方案不需要强迫任何人，只是简单地将是否生育的决定权交给女性。从他这个表述看得出来，其实就算在美国很多言论也是要很小心的。他这很明显指的是某些美国某些宗教氛围较重的州经常推出禁止女性堕胎的法律。关于这点我是认可的，还没有成型的胎儿是不应该作为一个具有完整权利的人来看待的，他需要借助母亲的身体才能存活。所以自然是否生育的权利，如何使用自己的身体的权利应该属于母亲。
人类智能对地球的另一大威胁就是人类新脑可以构建复杂的世界模型，理解世界的本质，并且设计制造复杂的工具。这使人类获得了前所未有的影响自然界的能力，但是这种能力仍然受到旧脑的驱使，而旧脑仍然保持着对基因存续有利的暴力攻击的行为。两者相结合就会非常的危险。比如当核武器落进了某个战争狂人的手里，他会毫不犹豫地使用它来毁灭自己的敌人甚至整个世界。

人类智能的第三大威胁其实和人工智能也差不多，因为虽然智能来源于新脑，也让人类更具远见，一定程度上能够克服旧脑的短视带来的威胁。但是新脑本身也是一个危险源，就是它通过学习和构建世界模型来运作，而它构建的世界模型如果是错误的就会带来危害，就像现在的大语言模型经常会胡说八道，给出错误的信息一样。作者这方面举的例子是那些质疑全球气候变化的人，不过我觉得这个例子举得实在是糟糕。作者一上来就一口咬定气候灾难是不可否认的事实，但是却没有给出任何的证据。倒是反方观点他倒是描述得非常清楚，就是气候灾难是政府推出的重新构建社会经济秩序的由头，是气候科学家们为了长期获取政府资金而编造的。这些怀疑完全是合理的，但是作者却把它们归为妄想和阴谋论。
不过还是看一看作者可取的一些可取的观点吧。其实这些错误的世界模型和语言的产生有很大的关系。在语言出现之前，大脑构建的模型都是基于我们能够看到和感受到的,由于语言的出现,人类大脑构建世界模型的能力得到了扩展。比如某个人居住在山区,从来没有走出去过。但是他的伙伴曾经看到过山外的世界,并将其描述给他听。比如翻过那座山就有一条大河了,虽然那个长期居住在山里的人从来没有翻过那座山,但在他的世界模型当中,山的那边有一条大河。而这个模型可能根本就错的。由于作者写这本书时，大语言AI模型还没有取得今天这样的突破，但作者已经意识到单凭语言也能构造世界模型。可以说在AI领域，先知先觉了。

---

Recently, I read a book called "A Thousand Brain Intelligence". The author's main point is that artificial intelligence is not scary, human intelligence is the most scary. Although there are some new and interesting points in the book, it feels like the author is carrying a lot of private goods. From his examples and arguments, he is a typical white left. He strongly believes that climate change is caused by human activity and will destroy the entire planet, a fact he often uses against his opponents. Let's see how he argues that human intelligence is more dangerous than artificial intelligence.
The author's theory of intelligence divides the human brain into old brain and new brain. The old brain has evolved over a long period of time, over billions of years. This process is completely driven by genes. The author was clearly influenced by Richard Dawkins' book The Selfish Gene. He believed that living things were nothing more than survival machines for genes, and that the basic function of living things was to survive and allow genes to continue in the process of reproduction. The function of the old brain evolved around this purpose. Because a gene is essentially a string of DNA molecules that only know how to copy, it has no feelings, no moral sense, and it can't feel pain. Therefore, it does not distinguish between good and evil of the survival strategy of the behavior, as long as it can ensure its replication and transmission is a good strategy. So the old brain is formed to allow us to have all kinds of emotions, to care for young children, to cooperate between different individuals and so on. These behaviors are conducive to the survival of genes, but they are also conducive to the survival of genes through resource looting, violent assault, and rape. While our old brain is unable to distinguish between good and evil of these means, only the neocortex of the new brain has mastered the complex model of the world, and can make more accurate and long-term predictions. Only then will we know that although some behaviors can bring benefits in the short term, in the long run they will destroy human living environment and social stability.
But because the old brain is the result of early evolution and controls the behavior of the body from the bottom up, it is difficult for humans to control it. Just like many people know that eating too many sweets and carbohydrates will damage health and increase the risk of cardiovascular and cerebrovascular diseases, but when a delicious cake is placed in front of you, most of them will eat it.

Another typical example is population control, which also shows the author's white left view. He believes that the infinite growth of population will eventually destroy the whole earth and family planning should be implemented. Because the reproductive instinct is driven by genes, it is also very difficult to overcome. However, the author says that humans have now invented a lot of contraceptive methods, which he explains as the new brain's instinct to use intelligence against the old brain. However, although the author's point of view is biased towards the white left, there is still a good point. He offered what he considered a viable solution to the population problem, and one that did not rely solely on the coercive power of the government, as had been the case with family planning in our country. He said the plan would not force anyone, but simply give women the decision about whether to have children. From his statement, we can see that even in the United States, many comments should be very careful. He was clearly referring to the fact that some of America's more religious states often introduce laws prohibiting women from having abortions. In this regard, I agree that the unborn child should not be regarded as a person with full rights, he needs to rely on the mother's body to survive. So the natural right to procreate and how to use her body should belong to the mother.
Another threat to the planet from human intelligence is that the new human brain can construct complex models of the world, understand the nature of the world, and design and manufacture complex tools. This gave humans an unprecedented ability to influence the natural world, but this ability was still driven by the old brain, which still maintained its violent aggressive behavior in favor of genetic survival. The combination of the two can be very dangerous. For example, if a nuclear weapon falls into the hands of a war maniac, he will not hesitate to use it to destroy his enemies and even the entire world.

The third threat to human intelligence is actually similar to artificial intelligence, because although intelligence comes from the new brain, it also makes humans more far-sighted and can overcome the threat of the short-sightedness of the old brain to a certain extent. But the new brain itself is also a source of danger, in that it operates by learning and building models of the world that can be harmful if they are wrong, just as today's big language models often talk nonsense and give wrong information. The example the author uses here is those who question global climate change, but I think it's a really bad one. The author begins by insisting that climate catastrophe is undeniable, but offers no evidence. The counter-argument, which he describes very clearly, is that climate disaster is a government-imposed excuse to reorder the socio-economic order, and that climate scientists have invented it in order to secure long-term government funding. These suspicions are entirely legitimate, but the authors dismiss them as delusions and conspiracy theories.
But let's take a look at some of the author's best points. In fact, these wrong models of the world have a lot to do with the production of language. Before language, the brain built models based on what we could see and feel, and thanks to language, the human brain's ability to build models of the world has expanded. For example, someone lives in the mountains and never goes out. But his companion had seen the world beyond the mountains and described it to him. If you go over that mountain, for example, there is a big river, although the man who lived in the mountain for a long time never went over the mountain, but in his model of the world, there is a big river beyond the mountain. And that model could be wrong. Since the author wrote this book at a time when large language AI models had not yet achieved the breakthroughs they do today, the author has already realized that language alone can construct world models. It can be said that in the field of AI, the prophet had the foresight.