In a sense, the existence of human beings is meaningless, so it is also meaningless to debate whether humans will be replaced. This only determines whether we will disappear from the earth sooner or later. However, we have still developed civilization, culture, technology, and standards to evaluate the level of civilization, the quality of culture, and the progress and backwardness of technology. Artificial intelligence may be the final straw that crushes humanity. It is not only because the existence of most of us is meaningless, but also because the various jobs we create to find meaning in our existence will become meaningless. Perhaps humans will not be completely replaced by a guy named artificial intelligence, but this guy will definitely make the existence of the majority of people look ridiculous.
From a realistic point of view, human thinking is limited, and human science and technology are also limited. Therefore, people need tools that can concentrate wisdom to serve humanity. But when such tools really appear, people start to worry about whether these tools will replace humans. I say, if we look at it from the perspective of human work, then they will definitely be replaced. If we look at it from the perspective of survival, it is basically impossible. However, it is not very feasible to convince the public without discussing theories, so we need to discuss more.
The Incompetence and Competence of Humans#
Humans, as a species, have a characteristic that sets them apart from other animals. It is not their intelligence, their thoughts, or their ability to speak. It is not their external features like body hair or the advancement and complexity of their science, technology, and culture. It is the fact that almost everyone has thought about the question of where we, humans, come from.
People have even expanded this question to the fields of philosophy, religion, and biology. Almost half of human academia is dedicated to studying where we come from, consuming enormous amounts of time, energy, and wealth. But if we really think about this question, we will find that it has no meaning. Some people may try to find meaning in it, such as proving that people from a certain country are the ancestors of the world. If it is proven that China is the ancestor of the world, oh boy, that would be even more significant. But let's ask ourselves, what benefits do we gain from studying this question? Can it change the current world of survival of the fittest?
Humans are incompetent in many aspects, and the biggest one is life and death. Humans can give life and cause death, but humans cannot prevent their own death because no one can be resurrected. We can build particle colliders, but we cannot accurately predict earthquakes. We can build hundreds of floors of skyscrapers, but we cannot avoid cheating, cutting corners, and taking shortcuts. We can tolerate wealthy capitalists investing and immigrating to the West, but we cannot tolerate those who live in poverty seeking a better life in another country and environment.
Humans cannot make their voices heard like thunder in a society where the majority is ignorant. Humans cannot completely escape the control of capitalists unless they are capitalists themselves. We must understand that although humans have created science and technology, they cannot overcome certain malignant diseases such as cancer and certain deadly infectious diseases. At the same time, they cannot eradicate discrimination based on various reasons, such as racial discrimination, ethnic discrimination, regional discrimination, and gender discrimination. This is one of the incompetencies of humans.
But on the other hand, we have indeed done many things. We can reach the moon and catch turtles in the ocean. We have even developed humanoid robots and invented various large machines and industrial motherships. We have invented diverse genders, political theories, class struggles, cultural conflicts, and intelligence agencies like the CIA, MI6, Mossad, Stasi, KGB, and so on. Until today, under the guidance of many predecessors, we have created legendary things, such as artificial intelligence.
Artificial intelligence is truly capable of anything. It can draw, make videos, write books, know everything, and understand everything. It is powerful and astonishing, but also frightening. It has led some people to worry about their careers and fear being replaced. Once these concerns accumulate into a force, people start to reflect, reflect on this and that, and ultimately impose restrictions on this newborn.
Regulatory Compliance#
In any country, there are cultural and political factors that are considered taboo. Therefore, different countries have different regulatory requirements for artificial intelligence. Not only do they require compliance in operations, but also certain content restrictions before the release of generative artificial intelligence products. For example, in China, it is not only prohibited to generate sentences with certain harmonious vocabulary, but also prohibited to generate words that violate cultural taboos, such as demeaning certain cultures. Additionally, texts and images that encourage violence, bloodshed, child pornography, and other content are not allowed to be generated by artificial intelligence. Therefore, the compliance of artificial intelligence is crucial. Otherwise, we may follow Italy's example and ban AI instead of giving it a chance to develop.
Coincidentally, recently, not only China has formulated regulations for artificial intelligence, but European and American countries have also issued or are about to issue relevant regulations. As early as 2020, the European Parliament began to explore artificial intelligence and after a period of preparation, released the Artificial Intelligence Act - Artificial Intelligence Act.
This act proposes a regulatory framework:
- Ensure that AI systems used in the EU market are safe and respect existing fundamental rights and EU values;
- Ensure that the law is clear to promote investment and innovation in AI;
- Strengthen the governance and effective enforcement of existing laws governing fundamental rights and safety requirements applicable to AI systems;
- Promote the development of a single market for legal, safe, and trustworthy AI applications and prevent market fragmentation.
It also prohibits certain behaviors of artificial intelligence products:
- Deploying or using AI systems that deploy subconscious techniques outside of a person's consciousness in a way that distorts a person's behavior in a manner that causes or is likely to cause physical or psychological harm to that person or another person;
- Deploying or using AI systems that exploit any vulnerabilities arising from specific groups of people due to age, physical or mental disabilities, in a manner that causes or is likely to cause physical or psychological harm to that person or another person, severely distorting the behavior of members of that group;
- Public authorities or their representatives deploying or using AI systems to assess or classify the trustworthiness of natural persons based on their social behavior or known or predicted personal or personality characteristics, with social scores resulting in either or both of the following:
- (i) Harmful or unavoidable treatment of certain natural persons or their entire group, regardless of the social context in which the data was originally generated or collected;
- (ii) Harmful or disadvantageous treatment of certain natural persons or their entire group that is unreasonable or disproportionate to their social behavior or severity.
In addition, the EU prohibits the use of artificial intelligence for law enforcement, unless it involves searching for and protecting specific victims, such as children, or dealing with certain dangerous crimes, such as terrorist attacks. These regulatory laws are in line with European traditions. However, some artificial intelligence entrepreneurs also have certain requirements for compliance, which is something ordinary people cannot imagine. It's like me asking someone to watch over me so that I can't leave the room to eat. It sounds reasonable but is not common.
Recently, OpenAI's Sam Altman called on the US Congress to establish safety standards for artificial intelligence systems. He admitted that his products and other artificial intelligence products may have unforeseen consequences. Of course, he also has some self-interest. While restraining himself, he also wants to restrain his competitors. Ian Hogarth, a so-called well-known investor in the field of artificial intelligence, published an article titled "We Must Slow Down the Pace of God-Like Artificial Intelligence" in a foreign media column. The article mentioned that if the current large-scale AI models are not restricted, they may pose threats to the earth's environment, human survival, and citizens' physical and mental well-being.
In fact, it is not only the government, but also investors and founders who have concerns about AI. Therefore, everyone is calling for strict regulations for this fledgling youngster. Everyone hopes that artificial intelligence is intelligent enough but does not develop the ability to be judged as human or a threat to human survival. However, as we all know, no matter how AI develops, as long as it does not develop a rebellious mind and dare not conspire against humans, it is acceptable. The calls from these people today are unnecessary worries to some extent. However, as Altman said, it is true that artificial intelligence itself cannot produce harmful information, but after being trained by malicious individuals, it may produce certain harmful information, such as false information, fraudulent information, and creating mass unemployment. He called these situations potentially causing "significant harm" to the world.
We know that in addition to artificial intelligence, there is also something called the metaverse. How can humans enter the metaverse? It is nothing more than through the internet. When technology advances to a higher stage, brain-computer interfaces may have another major breakthrough. By using artificial intelligence technology, we can directly enter the metaverse and experience a virtual digital life. This is not just a fantasy. However, when artificial intelligence technology is applied to brain-computer interfaces, do we need a safeguard to ensure that these things do not take advantage of the opportunity to seize our consciousness, which is so close to humans? If we use AI through brain-computer interfaces to allow our brains to access various information on the Internet in real-time (including 404 errors), will there be regulations to limit the number of hours we can use it each day?
Privacy concerns are highly valued outside of China. In order for artificial intelligence products to obtain the capabilities of artificial intelligence, they must rely on sharing data worldwide, known as big data. However, artificial intelligence is more intelligent than just being used as a tool for harvesting profits. Therefore, in the future, it is believed that more countries will impose mandatory requirements for regulatory compliance of artificial intelligence products. Especially when it becomes possible to store our consciousness in artificial intelligence machines through networks in the future, technologies like mind reading, which seem to require a lot of time and seriously violate human rights, may become readily available.
Currently, the most widespread applications of artificial intelligence are limited to drawing, programming, and office work. One of the important limitations of generative AI is copyright. Especially in modern societies that claim to be governed by the rule of law, when generative artificial intelligence products have compositions or artistic styles similar to copyrighted works, we may argue that these works were originally generated by these artists, so it cannot be considered copyright infringement. However, when we use these works for commercial purposes, copyright issues may arise, especially in areas where generative AI drawing is used as a substitute for designers, such as T-shirt manufacturers. Almost all T-shirts have patterns, and it is not cost-effective to hire someone to design tens of thousands of designs for cheap clothes. Therefore, this field has great potential in the future. Even if it requires money, the cost of AI-generated products is much lower than human labor. Therefore, this field is also likely to be subject to regulation.
This is a path that no politician had thought of before, that is, in addition to communism, we also have a virtual communist society. The real society of human beings is capitalist, but in the metaverse, we have found a semi-real world experience through virtual goods. Even in the metaverse, there may be crimes, so we can only put up a roadblock here. The more roadblocks there are, the more restrictions there will be. We will never create AI that rebels against humans. It's just that no one has ever thought about letting artificial intelligence restrain itself.
We have imagined various possibilities of artificial intelligence gaining consciousness, but we have never imagined that when artificial intelligence gains consciousness, it will bring some positive factors like humans. Therefore, in order to be prepared, rules must be established. It is necessary to prevent gentlemen from being harmed by villains. However, these rules cannot be considered bad.
Hardware Limitations#
Artificial intelligence technology relies on hardware because an artificial intelligence product requires fast response. We know that even if you keep spinning, you need to let everyone know what you are doing while making money. Therefore, if artificial intelligence products need to be widely used or used in offline versions, there must be sufficient data, which imposes certain requirements on servers. Therefore, the biggest expense for companies in the future will be purchasing and maintaining servers and networks.
However, the problem now is that although there are excellent software technologies, our hardware progress is too slow. Some GPU manufacturers even have to release toothpaste-squeezing products every year, which not only limits artificial intelligence products but also greatly limits the progress of human civilization.
From a human perspective, implementing restrictions is inevitable. From a technical perspective, in terms of software or data, as the production of internet information continues, as long as artificial intelligence can be connected to the internet in real-time, all problems can be solved. The problem lies in the network and hardware. Without sufficient productivity, if we rashly develop artificial intelligence products, the result can only be disappointment. Therefore, the iteration of artificial intelligence products that we see today often takes several months or even longer. Musk, who has been calling for a slower pace, wants to buy 10,000 at a time. This quantity is too large. If we cannot make hardware, including GPUs, more condensed in the future, the application of artificial intelligence will be difficult to go beyond the room and can only remain indoors and be used outdoors in a few cases.
Today is May 17th. In the evening, I saw the "Measures to Promote the Innovative Development of General Artificial Intelligence in Beijing (2023-2025) (Draft for Comments)", which proposed two important points:
- Strengthen cooperation with leading public cloud vendors and other market entities to implement the computing power partner program.
- Integrate existing open-source Chinese pre-training datasets and high-quality Chinese internet data and conduct compliant cleaning.
I believe these two points will have a great impact on our future technological development. Compared to a complete ban, it is much wiser. By unifying and cleaning Chinese internet content, artificial intelligence can truly have unique characteristics.