Will robots replace humans in the future?

After George & middot; devor, an American, created the first programmable robot in the world in 1954, people occasionally issued warnings that robots would surpass and eventually replace humans in the future. However, such “replacement warnings” are either from the imagination of science fiction writers or from the simple prediction of individual scholars, which lacks reasoning. This is also the case. At that time, the discussion about human beings in the domain of computer also focused on whether computers or artificial intelligence would have consciousness or intelligence similar to human beings. However, most scholars’ answers to this question are still qualitative. However, in recent years, the situation is obviously different. One difference is that the people who issue the “replacement warning” are not science fiction writers and individual scholars, but become many experts, including today’s world-famous scientists and technologists. For example, Hawking, a famous physicist, once warned: “due to the limitation of slow biological evolution, human beings cannot compete with machines and will be replaced. The development of all artificial intelligence may lead to the end of human beings.” Musk is a famous technologist. He also described artificial intelligence as “the greatest threat to human survival”, and compared the development of artificial intelligence to “summon demons”. Another difference is that the “replacement warning” has already been proved theoretically and proved by facts to a certain extent. Especially when people found that the intelligent machine alpha dog, which defeated the world champion of human go, could learn by itself, they had to marvel at the fact that artificial intelligence would develop more and more rapidly and surprise more and more. It is in this context that the discussion on whether robots in the future will replace human beings has been carried out with unprecedented enthusiasm in the academic community. Many scholars have participated in the discussion and expressed their opinions. < / P > < p > the reason why this “replacement warning” has become a hot issue attracting public attention lies in the fact that different answers to this question will determine that human society will adopt different attitudes and institutional arrangements towards the development of artificial intelligence technology: either actively promote it or have some restrictions. In addition, different answers also mean that we can judge from this that which of the discussions on artificial intelligence other than “replacing warnings” are meaningful now? What problems are meaningless because they will not appear? For example, if there is no possibility that robots can surpass and replace human beings in the future, there is no need to continue to study the artificial intelligence problem based on this “possibility”. Therefore, the discussion on “replacing warning” is not only necessary and important, but also should form a consensus and clear conclusion as soon as possible. Then, what kind of point of view will this clear conclusion be? This requires a comprehensive analysis and evaluation of various views in the discussion, and then make a final judgment. At present, scientists and philosophers are the main scholars who discuss “replacement warning”. They have formed two opposite answers to this question. One is the “substitution theory” which gives a positive answer, that is, artificial intelligence will surpass and replace human beings and become the new master of the earth in the future; the other is “no substitution theory” which gives a negative answer, that is, artificial intelligence will always be It is impossible to surpass and replace human beings. < p > < p > the theoretical basis for Hawking and musk to issue “replacement warning” is the AI technology development goal theory and AI technology singularity theory which are popular in the field of artificial intelligence in recent years. The theory of AI technology development goal is formed by the discussion of many scientists from L.G. Goode in the middle of the 20th century to Nick Bostrom. They divided the development of AI technology into three stages: weak artificial intelligence, strong artificial intelligence and super artificial intelligence, and took super artificial intelligence as the ultimate goal of AI technology development. Nowadays, the existing intelligent machines, whether they are non-human intelligent machines or humanoid intelligent machines, i.e. intelligent robots, such as alpha dogs in the United States, Sophia in Egypt, and Shibuya Mirai in Japan, all belong to the first stage of weak artificial intelligence, which can basically only complete a single task. The second stage of general artificial intelligence belongs to strong artificial intelligence, can have the same ability as human, can be competent for any work of human through learning, but has no creative ability and independent consciousness. The third stage of super artificial intelligence is even better. It not only has the creativity and self-consciousness of human beings, but also is more powerful than all human abilities. It can be called superhuman artificial intelligence. In his 1956 paper, Goodall first proposed the idea of super artificial intelligence, he said: “a super intelligent machine can be defined as a machine that is far superior to human beings in all intelligent activities, no matter how smart people are.”. Since machine design belongs to one of these intelligent activities, a super intelligent machine can certainly design better machines. There will undoubtedly be an “intelligent explosion”, leaving human intelligence far behind. As a result, the first super intelligent machine will become the last invention made by human beings – if it is obedient enough and willing to tell us how to control it. ” In Bostrom’s words, “a fully mature super intelligence can accomplish intelligent upgrading, strategic planning, social manipulation, hacking technology, technology research and development, economic production and other tasks, which is why scientists think that super intelligence will control human beings.”. < p > < p > different from Goodall’s assumption of super artificial intelligence, there are various ways to realize it. They come from different ideas and discussions of many experts, which can be summarized as follows. The first path is whole brain simulation or human brain replication. It directly scans the human brain, inputs the scanned raw data into the computer, and then outputs the neural computing structure in a powerful enough computer system to build a computer simulating human brain. The second path is to increase the complexity. “We can control the earth because our brains are much more complex than even the brains of even the smartest animals,” Bostrom points out in. Therefore, as long as we try to make the computer have enough “neurons” and the computer system is complex enough, the computer can reach or even surpass the consciousness of the human brain. The third path is functional imitation. The proposer of this path thinks that the human body has 11 kinds of system functions, and the bottom layer of each system is a microprocessor. Therefore, as long as the artificial intelligence system is constructed by imitating the 11 system functions and combining them together, the intelligent robot can reach the level of human intelligence, and then it can achieve transcendence. Marvin Minsky, who has made a great contribution to the development of artificial intelligence, said: “if we can make a robot as smart as a human being, we can also make a robot smarter than human beings.” The fourth path is the combination of machine and biology. The main idea is to implant human genes or imitative human genes into robots, so that robots not only have more powerful logical thinking ability than human beings, but also have the wisdom of desire, emotion and illogical thinking. For example, a South Korean scientist is studying how to give artificial chromosomes to robots to make them have sexual desire. The fifth path belongs to the essential construction. This is the way to “learn from Husserl’s” imagination changes freely “, and analyze the essence of” intelligence “intuitively The specific operation steps of this method are as follows: display various possible types of intelligence, and imagine various possible forms of intelligence from this as the starting point, and finally eliminate the accidental components of intelligence and find the essential elements of intelligence “, so as to use it to create super artificial intelligence. The sixth path is to build artificial intelligence based on quantum computing. With the development of quantum mechanics and the increasing understanding of quantum properties, many famous scientists “put forward the quantum hypothesis of human consciousness” and speculated that “the underlying mechanism of human intelligence is quantum effect”. Therefore, some people think that “artificial intelligence based on quantum computing” can become artificial intelligence beyond human ability, and this super artificial intelligence will become “post human” to replace human beings. The singularity theory of AI technology is put forward by ray & middot; Kurzweil, a world-famous American inventor and futurist. In his book published in 2005, he predicted that artificial intelligence will surpass human intelligence in 2045, and this moment is the “singularity” of AI technology. The prediction of singularity is based on Kurzweil’s “accelerated return law” which is inspired by Moore’s law. Moore’s Law means that the number of transistors and resistors integrated on a semiconductor chip will double every two years. The accelerated return Law means that “the analysis of the history of technology shows that technological change is exponential, contrary to the common sense” intuitive linear view “. So we’re not going to go through 100 years of progress in the 21st century – it’s going to be more like 20000 years of progress. “. Because of this, “within a few decades, machine intelligence will surpass human intelligence and lead to the advent of technological singularity.”. Since many of kuzweil’s previous predictions of science and technology are effective enough, especially his prediction that the completion time of the human genome project will be seven years according to the “law of accelerated return”, although it is at least half less than that predicted by everyone at that time, the final actual completion time proves that his prediction is accurate. Therefore, once his AI technology singularity theory is put forward, it is also necessary to With the wide recognition of many people in the academic circle, singularity universities can be set up to recruit students from all over the world. In addition to using the theory of super artificial intelligence and AI technology singularity as its own argument, the “substitution theory” sometimes refutes the existing or possible challenge views from the opposite side. One refutation is aimed at the view that “robots, as inorganic substances, cannot surpass organic substances”. The refutation is: “however, no one can prove that the mind must have something to do with the body. In terms of information processing and data output acquisition, robot has more advantages than human brain. One objection is: how can the emulator surpass the simulated? How can artificial intelligence surpass the simulated human? The refutation is: “there is no theorem that proves that the created things can not surpass the creator. On the contrary, there are many opposite examples in real life. For example, children are smarter than their parents, which is very typical.”. There seem to be more scholars who deny that robots will replace humans in the future, including scientists, philosophers and occasionally writers. There are also quite a lot of negative arguments given by them, which can be summarized into the following three aspects. < / P > < p > the first aspect can be called technical demonstration. This is to prove that artificial intelligence can not surpass human intelligence from the perspective of technology. There are four kinds of proofs, all of which are made by artificial intelligence experts. The first is the negation of AI technology singularity theory. There are three reasons for its negation. The first reason is that the singularity theorists “ignore the very important point: the increase of computing speed is not equal to the improvement of intelligence”. Because the improvement of intelligence depends on the improvement of deep learning algorithm, and computing