Datawhale dry goods
Sharers: Wu Qing, Dr. Liu Wei
" The rapid development of artificial intelligence has added a lot of uncertainty to the future, and the future world is even more unpredictable.
Since the advent of ChatGPT, it has set off a technological revolution on a global scale. The rapid development of artificial intelligence has added a lot of uncertainty to the future, and the future world is even more unpredictable.
On April 27th, Weicao Zhiku and Shengjing Yuanshi Co-Creation Exchange jointly hosted the Yuanshi Deep Research Salon entitled "The Unpredictable World Behind the ChatGPT Popularity Speed", and invited the director of Beiyou Human-Computer Interaction and Cognitive Engineering Laboratory, Cambridge University Visiting scholar Professor Liu Wei, executive chairman of the Information Society 50 Forum, Duan Yongchao, founding partner of Beijing Weicao Zhiku, brought you an in-depth discussion on artificial intelligence.
The following is a transcript of the dialogue between Wu Qing, the partner of Shengjing Network Union Group and the founder of Qilin Media, and Professor Liu Wei. It has been edited and shared with everyone. Welcome to read and forward.
Guest profile
Liu Wei, Doctor of Engineering, Beihang University, researcher at the School of Artificial Intelligence, Beijing University of Posts and Telecommunications, director of Beiyou Human-Computer Interaction and Cognitive Engineering Laboratory, visiting scholar at the University of Cambridge, expert in China-US Track II AI dialogue at the Strategy and Security Research Center of Tsinghua University, Former Chief Scientist of the Human-Machine Fusion Intelligence Group of the Science and Technology Committee, Distinguished Researcher of the State Key Laboratory of Media Convergence Production Technology and Systems, Vice Chairman of the Cognition and Behavior Committee of the Chinese Society of Dental Research, and a member of the Urban Brain Committee.
Research fields include human-computer fusion intelligence, cognitive engineering, human-machine environment system engineering, future situational awareness mode behavior analysis/prediction technology, etc. He is currently a Distinguished Expert of the Expert Committee of the China Information and Electronic Engineering Technology Development Center, a review expert of the National Natural Science Foundation of China, a member of the National Ergonomics Standardization Technical Committee, and a senior member of the Chinese Association for Artificial Intelligence.
Moderator profile
Wu Qing, partner of Shengjing Network Group, founder of Qilin Media, a media company focusing on reporting on Silicon Valley and Israeli technology elites. The producer and chief producer of the multinational media brands "Silicon Valley Boss", "Global Chain" and "Global Boss", and the author of the books "Blockchain Truth" and "Go All Out". Former investigative reporter for CCTV's "Focus Interview" and "News Investigation". Winner of many well-known international national awards. Winner of the 2017 Top Ten Brand Women in China and the Top 500 Chinese Brand Women in 2021. Guests of the Israel Knesset Science and Technology Committee, the National Prayer Breakfast of the United States, the promotion ambassador of the Israeli Ministry of Tourism, the Australian Chinese Film Festival, and the Chinese media friends of many Silicon Valley and Israeli venture capitalists.
Master of Business Administration from Zhongnan University of Finance and Economics, member of the "Financial Media Executives" project of PBC School of Finance, Tsinghua University.
Quick Facts
1. Talk about the development status and changes of artificial intelligence
1. Beiyou Human-Computer Interaction and Cognitive Engineering Laboratory mainly explores human-computer interaction
2. The nature of artificial intelligence has not changed in ten years
3. The inflection point of real artificial intelligence is the emergence of a new system
4. Realizing general artificial intelligence: the new three theories and the old three theories must be changed
5. Dealing with the side effects of intelligence requires joint efforts of all parties
2. The application of artificial intelligence in the military field
1. Two major applications of artificial intelligence in the early stages of the Russo-Ukrainian War: logistics support strike and command and control system
2. The application of artificial intelligence in the military faces six major problems
3. The US military uses artificial intelligence in two directions: machine learning and autonomous systems.
3. Talking about the human-machine fusion intelligence revolution
1. Human-machine fusion intelligence is not a simple brain-like
2. Hume's question is the key question of artificial intelligence
3. Cambridge: an important birthplace of artificial intelligence
4. Artificial Intelligence Revolution: A Thorough All-round Change
1. The human-machine intelligence revolution is a comprehensive revolution
2. Differences between Eastern and Western scientific and technological revolutions: Western computing, Eastern "calculation"
3. The integration of Eastern and Western civilizations has become a trend
5. Humanity has entered an era of no major changes in a century
1. What was the great age like?
2. Development direction of man-machine environment integration: man-machine environment system intelligence
3. Four keys to realize the intelligence of man-machine environment system: input, reasoning, output and feedback
01
Talk about the development status and changes of artificial intelligence
Wu Qing: Last December, I wanted to do an in-depth event on artificial intelligence and ChatGPT, but I think now is the best time, because most people, including myself, have gone from the initial fanaticism to calmness, Only by transitioning from a slightly distorted state to a more normal state now can we calm down and talk about some real issues.
On the first day of this series, I hope to give friends a large coordinate with a relatively wide field of vision. The time axis and space axis of this coordinate are long enough for you to open up your field of vision, find the Cape of Good Hope you observe in this coordinate, and find yourself Or the location of my company, this is my original intention for this series.
Before the conversation, I would like to reiterate today's theme, called the unpredictable world behind the popularity of ChatGPT. You may be a little confused. Why is it an unpredictable world? Because when I was preparing this series, I read a lot of materials, and the frequency of unpredictability and uncertainty is particularly high. I think "uncertainty" is very certain. Why do we study it? Because it is closely related to the fate of mankind, otherwise there would be no motivation and necessity for research. What will artificial intelligence bring? What kind of mentality should enterprises and investors adopt, especially what specific actions should they take to deal with it? This is the content of the first interview today.
The first person invited today is Mr. Liu Wei. He is a researcher at the School of Artificial Intelligence of Beijing University of Posts and Telecommunications, the director of Beiyou Human-Computer Interaction and Cognitive Engineering Laboratory, a visiting scholar at Cambridge University, and the China-U.S. Track II AI at the Strategy and Security Research Center of Tsinghua University. Conversation Expert. I know the word artificial interaction but I don't know why. In today's interview, the topic of human-machine fusion intelligence will take at least 25 minutes to discuss, and there will also be the topic of human-machine environment system engineering.
I have been looking forward to this dialogue for a long time. Please take five or six minutes to introduce the main issues that BUPT Human-Computer Interaction and Cognitive Engineering Laboratory is exploring.
Liu Wei: Okay, thank you Mr. Wu! Mr. Wu also introduced just now that I am the director of Beiyou Human-Computer Interaction and Cognitive Engineering Laboratory. Our laboratory mainly researches two areas. The first is the human-computer interaction above the neck, which mainly involves human and cockpit, and the ergonomics of human physiology. Measurements, including simulation, space capsule out of the cabin, and some interface design.
Second, we have written several monographs on human-computer interaction, the interface between physiology and physics of the human below the neck. The other is human-machine fusion intelligence, which mainly involves above the neck. Above the neck is human-computer interaction, which mainly involves human and cockpit simulation, human physiological ergonomics measurement, including behavioral simulation, and some interface design. If you want to know more about it, our WeChat official account has a detailed introduction.
Wu Qing: Well, speaking of your experience in Cambridge, which year did you go to Cambridge University as a visiting scholar? Back before 2013, right?
Liu Wei: Yes, from 2012 to 2013, a whole year.
Wu Qing: In 2013, there was also a wave of artificial intelligence represented by machine learning. I read your past WeChat official account articles. Ten years ago, what did you doubt? What's disturbing? Now a full ten years have passed, which is a very long time for technology. Has your initial suspicion changed?
Liu Wei: Still no change. Now that new algorithms have emerged, the essence of artificial intelligence has not changed. After returning to China in 2013, I communicated with friends at home and abroad and found a problem. Most of the artificial intelligence we do is automation, and it is not called artificial intelligence in the strict sense. Everyone’s expectations are relatively high.
Automation is relatively deterministic. Just now you mentioned determinism and uncertainty. In fact, the input aspects involved in automation are deterministic, the intermediate processing can be programmed, and the output can also be deterministic.
However, both ends of intelligence are uncertain, and only part of them can be programmed in the middle. Now artificial intelligence is more deterministic, and after ChatGPT came out this year, everyone feels that artificial intelligence has reached a new level. In fact, I feel that it is still continuing. The previous output based on statistics and probability is essentially to solve the problem of the input probability of the relevance of the next word. Therefore, the current artificial intelligence is still a kind of advanced automation, and it is not to mention the intelligence similar to human beings or higher intelligence than human beings that everyone expects.
Wu Qing: Your understanding of intelligence may be more accurate than most people. After the release of ChatGPT, many people can be said to be emotionally fanatical, while you are calmer. So from now on, when do you think there may be an inflection point, or a breakthrough point, a turning point? How many years will it take for artificial intelligence to truly become the artificial intelligence in your mind?
Liu Wei: Time should be based on the degree of perfection in shaping or building artificial intelligence tools. The first wave of artificial intelligence was driven by knowledge, including expert systems and symbolism. Although a lot of work has been done, there are problems because human experience or knowledge is not enough. It may cover all aspects, so a data-driven neural network system has emerged.
After entering the system, a combined artificial neural network is formed. This artificial neural network continues the development of the previous two artificial intelligences. Based on data-driven artificial intelligence, it involves RNN, which is a cyclic neural network. It uses short-term memory, intense Processing, as a very important way, is a way of processing information in series. Later, the long-short-term memory mechanism appeared, so that the shortcomings of short-term memory were supplemented by the advantages of long-term memory for neural network processing.
The most recent time is the model in the main framework, and the important modules in the model are based on statistics and probability plus hints. The important feature of the model is that the weights and processing methods of different inputs have changed from concatenation to processing, which has accelerated the speed. This has formed the GPT format based on the transformer model of artificial intelligence, which is very novel to everyone. In the next stage of development, we judge that it is a new form of intelligence based on context, situation, and situation, which may be the main development direction in the future.
Therefore, real intelligence is not only the processing method of the machine, but also involves the processing method of the complex human-machine environment system. It will take some time to look at the representation, and it needs to make a breakthrough in mathematics and have new insights into the new intelligent mechanism. know. In a word, a new system should be produced. This system may not be produced on the basis of GPT, not purely based on knowledge or data structure, or it should be called general artificial intelligence. It may be a new architecture. This framework is based on context and situational awareness. We are also looking forward to new breakthroughs in this area in our country. Thank you, Mr. Wu!
Wu Qing: Well, you think general artificial intelligence will definitely appear. I remember reading a book not long ago, "The World in 5000 Days" by Kevin Kelly. It should be mentioned in this book that general artificial intelligence will not appear. What's your take on general artificial intelligence? I feel that your point of view seems a bit different. I haven't gone deep into this aspect, nor do I understand it very well.
Liu Wei: Yes, I once met Kevin Kelly at the China National Convention Center.
Wu Qing: Is it 2014?
Liu Wei: Yes, he has been here.
Wu Qing: I also saw it at that meeting.
Liu Wei: Really, very good! There are similarities and differences between his point of view and mine. The difference is that he believes that general artificial intelligence may not appear, and that general intelligence is likely to appear within a certain period of time and in specific scenarios. General intelligence is not purely based on machines, but should be a comprehensive manifestation of the human-machine environment system, including not only technology, but also management, operation, and organizational systems, rather than individual intelligent products. I probably understand it this way .
Wu Qing: It is possible that your definitions of general intelligence are not exactly the same, so you have different judgments. But according to what you said, to achieve the general artificial intelligence you defined, the requirements are not simple.
Liu Wei: Yes, that's right.
Wu Qing: Certain conditions are also required, can this be understood?
Liu Wei: That's right, it can be understood that intelligence theory, control theory, information theory, system theory, and collaboration theory need to be refurbished. The new three theories and the old three theories will all undergo new changes.
For example, information theory, information theory only reflects the amount of information, which is measured in bits. In fact, information theory in future intelligence includes not only the quantity, but also the quality of information and whether the information is good or bad. In this way, there will be a new form of intelligence, Especially for people, great changes will take place.
Wu Qing: I understand. Let’s discuss the next question. I see that after you came back from Cambridge, you wrote two books, one of which mentioned the side effects of artificial intelligence. To be honest, I am more concerned about this issue. What do you think of artificial intelligence? What are the side effects? Why do you spend so much space on this issue in your book?
Liu Wei: Strictly speaking, artificial intelligence itself is neither good nor bad. It is a technology and a tool.
Wu Qing: Agreed, yes.
Liu Wei: Just like the atomic bomb, it involves good or bad users. Good people use it to benefit mankind, and bad people can harm mankind. Therefore, artificial intelligence has been abused a lot. When I was in the Cambridge Psychometric Laboratory, several PhDs participated in it. Cambridge Analytica, Cambridge Analytica later had some problems in the US election or other aspects. You can see that this will cause some social problems.
Similarly, telecommunications fraud, or some social crimes, especially data falsification, image falsification, and video falsification may cause many problems to society, so I am very worried that artificial intelligence will have adverse consequences if it is abused. This time, Musk and others also have this concern about the development of artificial intelligence after GPT4.0, which probably means this.
Wu Qing: I understand, indeed, especially in the few months since ChatGPT came out, I saw a lot of videos and images made by AI. The degree of artificial intelligence simulation is very surprising. In other ways, it is used by bad people, and the consequences are very dire. This is also a question that naturally came to my mind when I was planning this series and reading the materials. I especially agree with what you said earlier that the technology itself is neutral. The key is that this "seedling" is planted in human civilization. Are there any hidden dangers? If there are hidden dangers, coupled with technology, evil flowers may bloom.
Does the social problem of robots mentioned in your book belong to the same problem as the side effects of intelligence you mentioned just now? Or is it another level of problem?
Liu Wei: Quite a few of them are overlapped, and some of them are not. The social problem of robots also involves the problem of labor force. This is a social problem, not a crime problem. We all know that recently India's population has surpassed that of ours. India's population is particularly young, and its purchasing power and productivity will undergo new changes. Many people are also worried that our population is small and aging. question? Now, I understand that many people are more optimistic, why? Because we have robots, robots will replace a lot of labor in the future.
Of course, robots replace the labor force. If there are too many replacements, many people will lose their jobs. It is necessary for the government, society and people from all walks of life to jointly propose solutions.
02
Applications of artificial intelligence in military
Wu Qing: One of the fields you mentioned in the book, you may be very good at this field, but many people don’t understand it, is the use of artificial intelligence in the military. Anyway, among the experts I know in China, few seem to be discussing it.
Liu Wei: Yes.
Wu Qing: I am very interested in topics in this field. As a woman, I paid special attention to the Russo-Ukrainian War for the first time, and spent a lot of time researching and observing it. I would like to ask you, what did artificial intelligence do in this war, especially in the first three or four months? How do you evaluate the significance or value of these things done by artificial intelligence?
Liu Wei: I used to be similar to President Wu, and I didn’t know much about the military. Later, I participated in the Xiangshan Forum, which involved many friends from the military. Last year, I was also fortunate to participate in the online forum of the Xiangtan Forum, mainly American experts, Russian Experts, Indian experts and domestic AI experts. Among them, American experts asked Russian military experts how much artificial intelligence was used in the Russia-Ukraine war, and where was it used? I listened to it from the side, and it was very interesting. Russian experts said that the overall level of automation is relatively high, because they are fighting conventional mechanized and informationized wars.
For the application of artificial intelligence in the Russian-Ukrainian War, it is mainly in the logistics support strike and command and control system. In these two aspects, artificial intelligence technology has been used to do some work. There are a lot of image recognition technologies in reconnaissance. There is also one called Didi Fighting and Menu Fighting. Many Ukrainians ride electric motorcycles to the combat point after placing an order. In addition, on the reconnaissance side, there are mainly some smart sensors. Also in terms of attack, such as drones, there are a lot of artificial intelligence in drones. And cruise missiles also have some artificial intelligence, image recognition technology, and some speech synthesis. Probably this is the case.
We are very worried about the militarization of artificial intelligence. At present, artificial intelligence has many shortcomings. Especially if it is used in a bad way, it will cause social turmoil and war. In this regard, we hope that the military application of artificial intelligence technology will be restrained and constrained, so that it is conducive to peace.
During the Sino-U.S. dialogue and negotiations, American experts often raised this question, and they were also very concerned about the use of artificial intelligence in the military. The questions they raised probably included six aspects:
The first is the issue of data, which data can be used, which data cannot be used, how to prevent data from being misused, and the protection of personal privacy of data.
Second, it involves where artificial intelligence can be used and where it cannot be used, such as nuclear accusations. This aspect cannot be used because artificial intelligence technology is currently relatively fragile. If it is used in such accusations, there will be a situation that cannot be controlled by humans. .
Third, what is autonomy and how to define autonomous weapons? American experts define it in this way. The so-called autonomy is the quality and state of self-management. For swarm intelligence, autonomy is the degree of independence of its components. Others believe that autonomy may have other meanings, such as learning, adaptation, collaboration, etc.
Fourth, when it comes to the combination of people and equipment, whether people are in the environment and the system, on the system, or outside the system, many experts believe that people cannot be separated from the weapon system, and disastrous events will occur if they are separated from the weapon system.
Fifth, with regard to legal, moral and ethical issues, it involves a lot of internal values, so now we need to formulate some laws and some precautionary regulations to constrain.
Sixth, it involves testing and evaluation issues, which test indicators and evaluation methods can be used, which ones are not easy to use, and which ones cannot be used.
After some discussion on these six points, the United States seems to have issued a report last year, which means responsible artificial intelligence. I appreciate that and should have been responsible in the first place.
Wu Qing: Really.
Liu Wei: But being able to speak and act is the style of a leader. You can't just talk, you have to do it. We will wait and see.
Wu Qing: Yes, the word "responsible" sounds quite safe. I especially like to read articles by the scholar Liu Junning, especially on enlightenment. During the Russo-Ukrainian War, he also wrote an article called "Ominous Bingwu".
Liu Wei: Yes.
Wu Qing: I think of his article, and I also think of what you talked about a few minutes ago about sending orders to fight wars. I want to follow up a question, like smart sensors, drones, for attack, and for commanding the brain, I have heard some in the past. But how does dispatching orders to fight and speech synthesis work? It's not very clear yet, you can give me a quick scan of literacy in about a minute.
Liu Wei: Everyone knows Didi Taxi, you can send a message, and then a car will pick you up. In fact, sending a single to fight is to let the combat troops arrive at the designated combat location through training or through some command and control systems to carry out effective offense and defense, which probably means this. It is dispatch-based. It used to be centralized, but now it is distributed.
You can obtain the intelligence of the other party accurately. For example, Musk can see a lot of Russian armor, or a lot of Russian command systems, which can be passed on to the Ukrainian military. The Ukrainian military will carry out targeted attacks based on accurate information.
Wu Qing: Just now I looked at the data of the live broadcast, and more and more friends entered the live broadcast room. When I heard the professor talk about sending a single to fight, everyone may see my expression is a bit complicated. Speaking of the military intelligence of the US military, I have come into contact with a Jew who lives in Silicon Valley and was sent out by the 8200 troops of Israel. Unit 8200 is very powerful. Many Israelis who started businesses in Silicon Valley and Israel have been trained in Unit 8200.
This piece will not be expanded. I just want to ask you, what else do you know about the current military application of artificial intelligence in the US military? I wonder how far it has progressed now?
Liu Wei: Yes.
Wu Qing: Another question. In the future, intelligence may be more and more used in the military, because it is used in various fields, and it will inevitably be applied to military and national defense. What are the key factors to win intelligent wars in the future? Let's talk about these two issues again.
Liu Wei: Well, I also mentioned military intelligence in the book "Human-Machine Fusion". In fact, the US military uses artificial intelligence a lot, mainly in two directions, one is machine learning, and the other is autonomous systems.
Everyone knows that there is a famous case in artificial intelligence. Kasparov was defeated by Deep Blue. After seeing the power of Deep Blue, the United States launched a new plan called the Dark Green System. What kind of system is this? It is an auxiliary decision-making system. This system consists of three modules. The first module is called the crystal ball, the second module is called the lightning station, and the third module is called the commander assistant.
The Lightning Station fuses real-time data with past data, the crystal ball makes real-time predictions, and the commander's assistant recommends it to the commander. The commander selects the best one from the regulations or combat plans obtained from several machines for assistance. implement. This system is mainly based on computing. At that time, the computing power was not so large, and the modules were not mature enough, so the project was stopped later.
Now it seems that we are also working on a lot of auxiliary decision-making systems. There is an idea in these systems, which is how to combine basic theory and application engineering as mentioned just now, especially how to combine humans and machines. We are doing some human-computer interaction, human-computer fusion intelligence, etc. They are much ahead of us, relatively speaking, because their software and hardware are very good, and the hardware can be "stuck".
At present, everyone is exploring the integration of human-machine systems. The work in this area is very cumbersome, and it cannot be realized with a good product or a very good soldier. One plus one is not necessarily greater than one, and a person and a machine are not necessarily better than a person. They are also groping in this area.
03
Talking about Human-Machine Fusion Intelligence
Wu Qing: You have multiple research directions, but I think artificial intelligence ranks first, and it must be your special main direction. This can be expanded a little bit: how do you define human-machine fusion intelligence? What is the main idea conveyed in your book?
Liu Wei: Well, following the earliest topic, human-machine fusion intelligence is not exactly the same as artificial intelligence. In a sense, artificial intelligence is an advanced form of automation based on mathematical models or statistics and probability, because the output is certain, while intelligence is an uncertain input, partly programmable, partly non-programmable, The output does not have to be a deterministic system.
This uncertain system is often self-interested. What is the biggest feature of this system? It is not a simple brain-like brain, for example, a wolf boy, whose brain is also a human brain, but the growth environment is in a wolf den, and the habits and behaviors are the behavior of wolves, so we believe that a simple brain-like brain cannot be similar . Intelligent.
Wu Qing: Yes.
Liu Wei: Marx had a famous saying that man is a social animal and a product of society. Therefore, only in the social environment, in the process of social development, in the process of social communication, communication, and interaction, can genuine human intelligence be produced. Therefore, we believe that real intelligence is the product of the joint action of people, objects, and environmental systems, while machines are man-made objects, so sometimes it is simplified and called human-machine environment system intelligence.
Our research also involves the direction of human-machine environment system intelligence, involving some of the most basic issues, and requires everyone to do some exploration, and we have also done some work in this area. The first is Hume's question, which I also mentioned in this book Hume's question, why is it said that Hume's question is the key question of intelligence? On this question, he asked like this, whether we can derive subjective value from objective facts, that is, whether we can derive should from being. This question is very interesting.
The West believes that it cannot be realized, because being is objective and what should is subjective. If being is technology, then should is ethics. So it's two different things. But in the Eastern thought system, there is a very important relationship between the two. The Eastern people believe that the sky is healthy and the gentleman is constantly striving for self-improvement.
What does Tianxingjian mean? It is the objective laws such as universal gravitation and the theory of relativity that are at work. A gentleman must strive for self-improvement, which means that people's subjective moral ethics should be positive and should not be left behind. Therefore, in Eastern thinking, especially in ancient Chinese thinking, organically combining being and should forms a very good idea of the unity of heaven, earth and man. This idea plays a very important role in the development of the entire intelligence in the future.
In addition to Hume's question, there are two important directions: one is Hume's difference, which is to classify knowledge into key knowledge and factual knowledge. The key knowledge is equivalent to the mathematical knowledge that the sum of the interior angles of a triangle is 180 degrees. This kind of key, objective, and definite knowledge, the factual knowledge is about experience, and experience is uncertain and sudden. . Together, these two kinds of knowledge constitute the source of human experience and knowledge.
In addition, Hume also has another important idea, which is about the understanding of reason and sensibility. He thinks that reason is the slave of passion. What does it mean? That is, he believes that the source of all rational behavior lies in sensibility. For example, if it is raining outside, we need an umbrella to go out. On the surface, it seems to be a rational behavior, but in fact it is an emotional behavior. Why do we need an umbrella? It's because I'm afraid of getting wet, and it's uncomfortable after getting wet, so I have to take an umbrella to shelter from the rain.
Hume's question has had a great influence on Western philosophy, and Einstein and many Western philosophers, including Wittgenstein, Russell, and Kant, have been greatly influenced. When we talk about ChatGPT, we will say that ChatGPT is a large language model, which is the natural language processing of NLP. In fact, we can see from the two great works of Wittgenstein who were most influenced by Hume's question, because Wittgen Stein is a great philosopher who studies natural language processing and language analysis. In his first book, he wrote a Treatise on Logic and Philosophy. He equated the world with language. He believed that the world consists of three parts .
The most basic unit is the object, there are various objects, such as drinking water, the water cup is an object, this object and the table constitute a state of affairs, the state of affairs and the room constitute another state of affairs, and the whole world is formed between the state of affairs and the state of affairs Facts, which constitute events, correspond to the names corresponding to objects in the language. This water glass is a name, and the basic proposition corresponding to this state of affairs is that the water glass is placed on the table, the basic proposition is the water glass and the table, what constitutes the table and the house, the proposition is formed between the two states of affairs, and the proposition constitutes the language. Therefore, he believes that the whole world can be described in language, and now GPT also has this idea, thinking that language is intelligence, text, and all artificial intelligence, but it is not.
Later, Wittgenstein discovered a very important problem, that is, the world and language are somewhat equivalent, but language cannot reflect the whole world, language is language, thought is thought, and human thought and language cannot be equated. Therefore, he wrote "Philosophical Research", in which he clearly put forward two basic points of view. The first one is that the most important thing in a world language is not only representation, but also use. Only when the physical world is related can it not only have syntax, but also semantics.
In addition, there are more important representations. He believes that in addition to structured, digital, and very good homogeneous things in the world, there are many volitional, unstructured things called non-family similarities. make up the complexity of the world. Therefore, the world not only has the existence of wind, horse and cattle, but also the existence of connections between wind, horse and cattle.
Therefore, Wittgenstein and his friend Turing had a discussion at that time. Turing was inspired by him. Turing always believed that mathematics was a discovery, and Wittgenstein believed that mathematics was an invention. Where are the main representations of Wittgenstein's thought? He thinks that mathematics is an exploration, which is to explore things that are unknown, while Turing thinks that mathematics is a tool, something that already exists, so they have been arguing. The three-point land gradually developed, forming a raging fire, which slowly extended to Europe and the United States.
So I personally think that the development of artificial intelligence is very interesting, and what is even more interesting is that the hardware of artificial intelligence is related to Cambridge, and the software is also related to Cambridge.
Wu Qing: What is it?
Liu Wei: The world's first great programmer, Ada's father, is Byron of Cambridge. Byron is a great poet. He belongs to Jiansan, so Cambridge is a very interesting place. It has both hardware and the earliest machinery. The invention of the extension machine is also in Cambridge. Microsoft has a research institute in Cambridge, and the scale is quite large. It is a very interesting and long story, probably like this.
Wu Qing: Your book is really good-looking. In the words of today’s young people, it’s particularly good, because it contains many keywords that I am interested in, such as sensibility, rationality, the East, and the West. I think this is your book as a Scholars, from the perspective of natural comparisons at home and abroad, I just talked about Wittgenstein. About ten years ago, when I was still on CCTV, I read Wittgenstein’s philosophy of language, but I didn’t quite understand it. His biography also talked about the story between him and Turing. Later, he was influenced by him and read some of Russell's books, as well as Hayek. They seem to be cousins.
Liu Wei: Yes, Wittgenstein's cousin is Hayek.
Wu Qing: Wittgenstein is very interesting, he is a very innocent person. Before he died, he said that he lived a very good life. I think this sentence has a great influence on me.
Liu Wei: That's right, I went to Wittgenstein's cemetery three times, which is in the Cambridge Cemetery. Every year, many worshipers from all over the world visit the cemetery for very interesting and common purposes.
Wu Qing: Professor, you are a very interesting person. I think of another person. It is Mr. Zhang Shoucheng, the late Stanford professor. He was 16 years old and could not continue studying abroad. The power of action.
Liu Wei: I dare not, eh!
Wu Qing: You all like this, it’s so interesting.
04
AI Revolution:
A radical change across the board
Wu Qing: You also mentioned a problem in your book, which is a huge...I don’t know if there is such an adjective, anyway, a revolution sweeps human and machine intelligence. What kind of revolution did you refer to when you wrote the book? ?
Liu Wei: That’s right. Now the entire revolution is a scientific, philosophical, humanistic, and artistic revolution. Everyone knows that artificial intelligence like GPT can be faked, and can produce many subversive jobs that did not exist before. So this revolution It will be deeply imprinted in different disciplines and fields in all walks of life, including the human way of thinking and cognition of the world will also undergo great changes.
So the revolution I mentioned is a bit like the Renaissance or the Enlightenment, or even bigger, because it was just the beginning, and now humans are not only going to revolution on the earth, but even go to space, and Musk’s SpaceX rocket has also begun to launch In a sense, his failure is a kind of success, which will affect the development of the whole human being. After a few years or in the near future, human beings will laugh at themselves, why are they still fighting in such a small place? , Why not go to space to do bigger things, probably like this.
Wu Qing: 100 billion galaxies.
Liu Wei: Yes, it’s amazing!
Wu Qing: What I want to say is that it is not only a revolution in technology, but also a comprehensive revolution in all fields of humanities. The technological revolution is actually difficult to see clearly, and the boundaries are not easy to define. I think our imagination may or may not be able to imagine the intricate influence of each other.
Liu Wei: Yes.
Wu Qing: I wonder if you were aware of this when you wrote the book?
Liu Wei: I think there is still something that can be described. This revolution is not just about technology. Technology is a product of the West, and it is the relationship between people and things. More importantly, it will gradually incorporate many Eastern thoughts, including those from India, China, and the Middle East. India is a very special country, mainly religious, which is the relationship between man and god, while China is mainly the relationship between man and man, man and environment, so this kind of cultural integration and collision of civilizations will appear. There are many secrets that could not be discovered before, which will affect the trend of the whole world.
Some politicians say that the clash of civilizations can be reversed in some respects. If civilizations continue to merge, then human development will undergo qualitative changes. Sometimes they joke that Western technology is mainly developed in the form of calculation, while many ideas in the East are calculated. thought of. Calculation here is not a derogatory term, but a neutral term, which involves calculation, operation, macroscopic and overall perspectives.
The West is reductionist, cutting everything into small pieces, such as atoms, molecules, and quarks. In this way, reductionism can be combined with systemism or holism, which may produce revolutionary changes. For example, in quantum physics, it will be reflected in quantum physics. This is Schrödinger's cat, which can live and die at the same time.
In fact, ancient China has already had this kind of thinking, which is the coexistence of life and death, and the yin and yang and Tai Chi are in one system. I feel that many Western scientists, including some Nobel Prize winners, are slowly absorbing some Eastern cultures, and the two sides should complement each other.
Many people who are doing artificial intelligence in China now have a background in science and engineering, and they have studied Western mathematics, physics, and chemistry. There are two sculptures in Cambridge that make me very emotional. There is a vegetable market there, which claims to be thousands of years old. Next to the vegetable market is a statue of Thales. Thales is the originator of Western science and philosophy. There is another very important one. In the back garden of the garden, there is a sculpture of Confucius, which is a very important symbol of Eastern thought.
Wu Qing: Yes.
Liu Wei: So I can clearly feel in Cambridge that the East and the West are merging, and everyone is learning from each other's strengths to complement each other, so the great revolution you mentioned just now is happening in China, Cambridge, the United States, and even other countries. I probably think so.
Wu Qing: I understand, this is very interesting. In the official account of your laboratory, you also have this feeling, and you will talk about many Western things, but there are also many Eastern ones. It is really emotional and rational.
05
Humanity has entered an era of no major changes in a century
Wu Qing: The great era you described just now, in fact, everyone seems to have a lot of emotional descriptions of this great era, and the titles of some books refer to the great changes that have not been seen in many years. But what kind of era this era is, in fact, the general public is not clear, and I am not particularly clear, so what kind of era do you think it might be? Is it as beautiful as the previous Axial Age? Or another one?
Liu Wei: Yes, I feel that there are both. None of the civilizations appeared suddenly, they are continuous. Our country has a very prosperous civilization in a long time ago. In fact, after World War II, many Western civilizations also appeared, including After the Renaissance, a very prosperous civilization era of science and technology, humanities, and art emerged. So far, economically speaking, the East and the West are getting closer and closer to each other. When people are full, they must eat well.
Wu Qing: Yes.
Liu Wei: When eating well, what do you think about? In jargon, it is called functional science, which means safety, health, and comfort. In fact, many ideas of the East and the West are now being integrated, and the speed of integration is also accelerating. Especially now that many children can speak English. I believe that in the near future, many foreign children The baby can also speak Chinese, and it is normal to learn Chinese, which speeds up the integration and symbiosis between the two parties. With the emergence of technological leaders like Musk, the narrowness of the earth is gradually opened up, and a community of shared future for mankind is gradually formed, which should be the case.
There may be some plagues and wars in the future, which are not enough to destroy mankind, but will affect the progress of mankind. If there are many problems in the universe in the future, which will cause danger to human beings, including temperature rise and climate change, human beings must unite to go further. So I feel that as politics, economy, and culture are gradually approaching, there may be major changes, coupled with the emergence of some external influences and threats, human beings may gradually become better.
Wu Qing: I agree, and I am looking forward to it. Like what you just said, when you talked about various aspects of the environment, I remember five or six years ago, when I was in Silicon Valley, I interviewed a talented and beautiful Canadian scientist. Relatively large environmental problems will occur around 2050. After listening to what you said just now, I think you must read these two books. I only bought one of them, and I think I should buy the other one quickly. I read them during this holiday. These things are really interesting.
Liu Wei: Thank you.
Wu Qing: I think at this point, the purpose of the first episode of the artificial intelligence series has been achieved. In the next few episodes, I hope to talk about investing, and talk about very specific issues in business. I very much hope that the discussion with you and Teacher Duan today will give everyone a broad perspective and coordinates.
Liu Wei: OK.
Wu Qing: Only in this way, I think it is really good to find your own coordinates and your own position when you open your mind at the beginning.
Finally, I would like to talk to you about a question. The topic just mentioned is your main research direction, and it is also a topic mentioned repeatedly in the two books, that is, human-computer integration. So what is the future development direction of human-machine fusion intelligence? Or to put it another way, what kind of beautiful path could it be? To achieve this beautiful path, the road must be tortuous. Generally speaking, what problems may be very critical?
Liu Wei: Okay, thank you Mr. Wu. It is roughly like this. The development direction of human-machine environment integration is human-machine environment system intelligence. It should be in this direction, including four parts.
First, in the future human environment system, the most important thing is the input end, not just the data. I especially want to emphasize that the real intelligence is not big data, but a lot of small data, such as whether children learn and grow . So much data. In addition, data, algorithms, and computing power are just one aspect and cannot represent all aspects of the human-machine environment system. Especially at the input end, in addition to data, there are also subjective information, knowledge, and experience. A very important bottleneck for the development of the future human-machine environment.
The second one is about the stage of reasoning and information processing. This stage involves axiomatic reasoning and non-axiomatic reasoning. What I want to share with you is that I personally feel that people often regard intelligence as mathematics and mathematics as logic. , this is biased, because mathematics is not logic. Mathematics is a logical system based on axioms, such as Euclidean geometry, which is a logical system based on the five axioms. If there are no five axioms, mathematics will be fall apart. Intelligence contains both logic and illogical, both rationality and sensibility. This is intelligence. In the human-machine environment, perceptual problems must be solved.
The third is the output stage, which involves two basic issues. One is the logical output decision of the machine, and the other is the decision of human intuition. How to integrate these two decisions? Now it seems that we still need to work together to think. However, in this logical decision-making, humans cannot keep up with machines. Looking at AlphaGo and Deep Blue, we know that humans are no match for machines at all. But human beings have advantages in grasping the direction and handling morality and ethics.
The fourth issue involves feedback, which is how to integrate machine feedback and human reflection. So far, there is no good solution. Human reflection is more complicated, involving not only fact-based feedback, but also the value in Hume’s question sexual feedback. This combination of value and facts, at present, needs to be explored in depth. Therefore, we believe that the development of human-computer environment system intelligence requires everyone to work together and work together in many aspects.
Wu Qing: Okay, thank you professor, your timing is too accurate.
Dear friends, you are watching the live broadcast of the 16th session of the Round Rock Deep Research Salon co-sponsored by Weicao Zhiku and Shengjing Jiacheng Round Rock Co-Creation Exchange. You just discussed with Professor Liu Wei about human-machine fusion intelligence and other related topics. Really made me very teachable. I especially like to chat with experts and scholars with an international perspective, such as Professor Liu Wei and Teacher Duan, because they have rigorous academic research as scholars and can correct a lot of understanding of concepts for us. Thank you very much, professor, and once again recommend to you two books about human-computer fusion intelligence by the professor. I also recommend the official account created by the professor's team. It should be the name of your laboratory, right?
Liu Wei: Yes, Human-Machine and Cognition Lab, thank you Mr. Wu.
Wu Qing: This account is very good. Everyone will have a broader view of artificial intelligence in it. It is not about technology, but about many aspects, such as Eastern and Western philosophies, perceptual and rational. more open. The purpose of our first artificial intelligence event today is to open our horizons. Thanks again, professor, I hope to have the opportunity to do interviews with you and your team in the future. When will you publish a new book and what's new, don't forget to tell me, thank you!
Liu Wei: Okay, thank you Mr. Wu, goodbye!
" Like " three times together ↓