Ai smells dry! When will it be possible to conquer humanity?

“Comprehensive artificial intelligence may mean the end of humanity. Machines can start on their own and automatically redesign themselves at a faster rate. Due to the long course of biological evolution, humans cannot compete with them. Replaced." - Stephen Hawking The "Terrain" of "The Terminator" almost destroyed humanity. The concept of artificial intelligence comes from a series of panic attacks. When people first notice these theoretical people and present them to the public, they are actually the creators of some movies. I wonder if you are still a movie. Skynet in The Terminator has the impression that there are many kinds of artificial intelligence concepts such as this, and the concept of artificial intelligence may be another film subject after the alien intelligence species. AI is feared by many people like tigers But are these hypotheses truly groundless? In fact, it is not always the case. Since it can be repeatedly mentioned and supported by many people, it also shows that artificial intelligence is actually a potential threat to human beings. The AlphaGo defeated the professional chess player. It was not news any longer. AlphaGo Zero recently exploded AlphaGo with an absolute advantage of 100:0. This is in the field of artificial intelligence. It caused a series of sensations. What is the difference between Alpha and Alpha dogs? Alpha Dollar Fighting Alpha Dog The Alpha Dollar does not depend on any human record, does not refer to any human knowledge, and is entirely based on self-enhanced learning. In three days, it can personally broadcast 4.9 million chess games. Take 100:0 100 wins. Previously, the Alpha dog had spent several months studying the human thirty million chess game to defeat humanity. Alpha is no longer limited by human cognition. It can discover new knowledge and develop new strategies. It makes deep learning more convenient and feasible for complex decisions. There is no doubt that the emergence of such a "monster" once again triggered the "strong artificial intelligence" will replace the panic of human speech. Let's leave it alone for a while (later returning). First of all, let's talk about some of the artificial intelligence that we can come into contact with in our lives. Take a simple example. “Siri” is one of them. Artificial intelligence at this stage is Most of our lives have appeared as an "audio assistant" and are being used more in smart speakers and televisions. In other words, the current artificial intelligence is not as terrible as Skynet. It can also help you change programs and tap songs. At the current stage, artificial intelligence can't even allow you to completely throw away the remote control. That's it. Siri is also an "artificial mental retardation" So can these “silly guys” evolve the wisdom of conquering humanity? The answer is of course possible, but the time has not arrived yet. How do they conquer humans? You may wish to take a look at the blueprint for the era of highly-developed artificial intelligence compiled by the author. From the depth of the transition from AI to dominance, AI gradually conquered humanity, a common topic, and people who had conducted countless conjectures and obscenities thought that this was a “path to necessity”. Why did you say that? The reason is simple because of the demand. You know, AI is first and foremost a tool for mankind, and this tool will eventually grow to the extent that you can't give it up. It can help you in every aspect of life, astronomy, climate, medicine, and make life better. It Can bring great economic value. Even your map navigation is now a machine learning program is helping. So AI will exist and develop. Mature AI will serve you like "Jarvis" When it comes to your life, then a surprise comes in. A mature AI will blend into your life. Imagine an AI that can monitor your refrigerator. It knows what you like to eat. Afterwards, you will automatically place an order on the Internet to help you pay for electricity, water, and network fees. Later, you will be informed of your deductions and even help you plan the most favorable repayment method. This kind of service is so intimate. There is really no reason to reject it, and at this point, the so-called needs are transformed into dependencies. Are you using AI to avoid such road conditions? At the same time, the operation of AI will be more and more stable, because almost all technology companies are making efforts to it, and continuous R&D investment will make it more and more perfect. Will it not reach the stage of almost no error? The answer is inevitable. A previous AI combat example has proved this point. After AI studied 2 million cases of breast cancer, the diagnostic accuracy rate for this condition reached 98%. This correct rate is far beyond humanity because human beings cannot go. Learning 2 million cases, so it seems sooner or later that the hospital is occupied by artificial intelligence. AI can also help people diagnose cases When the AI ​​technology matures, after the combination of deep learning and deep learning? Its data learning ability will exceed 10 million, 100 million, then it will not go wrong, and a system that is never wrong and highly dependent, you will close it? Unconsciously, the basic aspects of society, your food, shelter, and livelihood have all been taken over by AI. This is the so-called process from being inseparable to the transfer of dominance, and the reason why AI is not yet a threat to humans is because it is still a “stupid guy” at this stage. Although it is powerful, it has tasks and functions. But it is simple. For example Alpha Dog, you might make him play chess very hard, but to fight a landlord is not enough. The next step of AI is the multi-tasking and even the implementation of fuzzy tasks. The stable operation of AI in various fields is implemented, but the world is one. For example, the relationship between weather and traffic leads to the whole body. If the weather is not good, the traffic will be affected and the road will be traffic jam. According to the severity of the natural disaster, the traffic AI will decide whether to close the highway, railway or flight. Under this premise, the traffic AI and weather AI will be combined together. Due to the intelligent arrangement of AI, before the weather deteriorated, the logistics system transported operations in advance and the vehicles escaped the congestion. People on business trips arranged their own trips more effectively and were more successful and winless. Typhoon strikes, mature AI will avoid risks and reduce losses But AI has been in charge of the meteorology and transportation department, because it is so easy to use, so we give it more authority. This is the process of transition of management rights. Since then AI has started from "weak artificial intelligence" to "strong artificial intelligence. "development of. The combination of weather and traffic has a great influence on the air quality of the city, so PM2.5, ozone and other air quality management functions are also linked together. Air pollution also affects the respiratory tract, cardiovascular and cerebrovascular diseases, and can lead to fluctuations in morbidity. Therefore, AI is involved in the medical disease prevention and control department. Emergency accidents and traffic are also related to each other and are large and complex. Strong artificial intelligence will eventually appear Finally, after continuous integration, there may be a monopolistic giant AI that will horizontally manage all the important fields related to humans. Since then, "strong artificial intelligence" has finally emerged that people fear. Since then, mankind has become a weak baby, and AI has become a strong nanny. All research and operation of the machine must be carried out under the conditions of management and regulation. There are relevant laws to regulate. These laws and regulations are like the parents of babies, parents, and nannies who do not dare to behave strangely. But will parents be beaten? Considering the financial crisis once in the past, all legal departments and banks were conscientious, but disasters washed away all macro-management in a systematic way. Will humans eventually become "little babies"? To give a simple example of the "strong artificial intelligence" hypothesis, a paper mill's AI has almost infinite power, it will use all resources, plant trees and make paper, repeat such actions indefinitely, just like "skynet" if It feels that humanity has hindered this task of papermaking. Will it choose to eliminate humanity? At this point, you will find that the motivation of AI is still as simple as the beginning, and the qualitative change is the transfer of authority. It's like a baby who knows nothing about the world, and you tell him that there is no law in the world. If you want to get something, he is weak and may not be enough. But if he happens to be a superman, it will be terrible. . All these are just some conjectures. It is not impossible to become a reality. Right is an overview of the current AI panic attacks. Now, we are looking back to reality. What is the most advanced artificial intelligence at this stage? How far does it want to conquer humans? Only 6 years old? Is it too early for artificial intelligence to be the boss? AI has been proposed since the 1940s. It has been developing for about 70 years. Do you want to know how smart it is? More and more people are beginning to advocate the artificial intelligence threat theory. At the same time, some people have begun to show concern for this issue. Some scientists who learned to dominate the game also started this research. They proposed to establish a standard intelligent model for artificial intelligence systems. In fact, it is a scoring mechanism. Is the scoring system for the robot objective and fair? In order to better analyze the intelligence level of AI, humans and other agents, they divided knowledge acquisition, mastering, innovation, and feedback into 15 small categories to evaluate AI and human intelligence from more dimensions. These 15 sub-categories are the recognition and output of images, text, and sound, common sense, calculation, translation, arrangement, creation, selection, guessing, discovery, and so on. But does this really measure AI's "IQ"? We can see that many of these measurement factors are indicators of perceptual analysis. However, AI's strengths are not these. To measure AI with human intelligence standards is obviously not objective enough, according to their conclusions. The best AI system is Google’s AI Alpha Dog, and its “IQ” can only be reluctant to compete with 6-year-old human children. It is too early to mention creation, but its search and learning ability is beyond the reach of human beings. In the author's opinion, for artificial intelligence, it may not be possible to talk about the perceptual aspects of creativeness. Therefore, this report does not produce much convincing. Alpha dog can win Ke Jie, 6-year-old. Can children? However, six-year-old children can create, Alpha dog can not, which is the difference between the machine and human beings, is the difference between absolute sensitivity and absolute rationality, and therefore can not be confused. However, they proposed a concept of “grading” for artificial intelligence. This, in my opinion, also has certain reference value. They divide AI into 7 levels, even if it is an indicator that artificial intelligence needs to reach in different stages. The 0th level system has its basic characteristics in theory, but there is no such artificial intelligence system in reality. The first level system is characterized by its inability to interact with human testers. The basic characteristics of the Level 2 system are its ability to interact with human testers, with controllers and memory, but the system's internal knowledge base cannot grow. The basic characteristic of the Level 3 system is that in addition to the features of the Level 2 system, the programs or data contained in the controller and memory can be upgraded or added without networking. The basic characteristics of the 4th level system, in addition to the features of the 3rd level system, are that the most important thing is to share information and knowledge with other intelligent systems through the network. The basic characteristics of the 5th level system are the ability to innovate and create, identify and identify the value of innovation and creation to human beings, and apply the results created by innovation to human development. The basic characteristic of the 6th level system is that as time advances and tends to be infinite, it will continue to innovate and create intelligent systems that generate new knowledge. Its input and output capabilities, and the ability to grasp and use knowledge will also converge. Infinity, according to Christianity's definition of God “omniscience and all-powerfulness”, we can see that in the case of continuous innovation, creation, and continuous accumulation of knowledge by intelligent systems, intelligent systems represented by humans in sufficient time will eventually realize “all-knowing and all-powerful”. status. Alpha dog is in the third system characteristic level, but Alpha element has the existence of some fourth-level characteristics, and all this starts from "neural network" and "deep learning". Keywords: neural network, deep learning As early as 1943, the embryonic form of neural networks has emerged. In the field of machine learning and cognition, the concept of building a neural network is inspired by the operation of human neural network functions. Many complex applications (such as pattern recognition, automatic control) and advanced models (such as deep learning) are based on it. It is the underlying model of artificial intelligence. Although concepts such as deep learning are relatively new, they are still based on mathematical theory that dates back to 1943. Neural network is a neuron transmission model that imitates the human brain There is no strict formal definition of neural networks. Its basic characteristics are trying to imitate the mode of transmission and processing of information between neurons in the brain. These may be somewhat esoteric here, but in fact, for humans, it is a process of information collection and screening. In order for this machine to realize this process, it takes millions to millions of operations. From the computational mode of the neural network to the process of “deep learning”, it is the case that mankind transports computing blueprints and masses to machines. Originally decided to screen and re-screen to draw the conclusion of the process, for example, Alpha Dog, after learning 30 million sets of chess game, the amount of calculations per time is very large, which is deep learning. Massive chess game learning is the magic weapon for Alpha dog The Alpha element of victorious battles, mentioned "reinforcing learning", has the ability to self-defense. Now is an unprecedented era full of computing power. The computing power of any smart phone on hand is certainly a boost. one. But in fact, the real key is from the concept of "information sharing", so it has the fourth-level factors. The seemingly powerful AI is still a "stupid guy" Can a seemingly powerful Alpha element be applied to other areas of life? Of course, the answer is no. We do not say that the cost of computing does not say that the machine wants to understand human intentions. It is not a good game to be able to do well. This is not to say that Go is not complicated enough, but is involved in life. The scene is diversified. The conceptual embryonic form of the neural network determines that it is the body characteristic of analysis and screening, and infusion of too many fields of knowledge can confuse the screening results. Take the example of "artificial intelligence voice assistance" on TV. If there is music with the same name as the movie, only searching for the title of the title will not be able to analyze the purpose of the user. Therefore, "I want to listen" or "I think "Looking at" the linked vocabulary, but also can not be a vocabulary not in the action vocabulary, replaced by "I think chanting" is not alright, simply analyze the information in the screening voice, it has been stretched, really into life after The problems AI faces are far from simple. Written at the end: Develop three aspects of artificial intelligence: "resolving the brain, imitating the brain, and surpassing the brain." Where is the current technology achieved? The neural network may be a preliminary action for resolving the brain, but for the second step, there may be a large distance on the emotional level. At present, it is still too early to surpass humanity and even dominate humanity. Artificial intelligence is a theory that allows humans to recognize the new concept of “robots,” not to complete governance walks. Those are just machines with better balance. The development of AI may be the right way, but it is also the most terrible. Yes, we may not need a big player who can run and jump. Instead, we need an all-embracing housekeeper. As far as the stewards are anti-guest, it is a matter of the depth of the transfer of the dominant power. There are policies to countermeasures, and there is a panic for artificial intelligence. Speech, implementation may be a matter of time, but this time may still take a long time. Smart TV/box information can focus on smart TV information network sofa butler (http://), China's influential TV box and smart TV website, providing information, communication, TV boxes, smart TVs, smart TV software, etc. Answering questions.

This entry was posted in on