Speech at Google: quantum computing, artificial intelligence and block chain

This article comes from: Phoenix network technology, the Star daily is authorized to forward.

Zhang Shousheng, Chinese American physicists, academicians of the American Academy of Sciences, academician of the National Academy of Sciences, and the foreign academician of the Chinese Academy of Sciences. He is now a professor of J.G. Jackson and C.J. Wood at Stanford University, and founded the investment in the frontiers of science and technology in the United States and China.

This article is compiled by Professor Zhang Shousheng in a keynote speech at Google.

The following is the full text of the speech:

Thank you for coming. I was very happy to come to Google, and it was a great honor to be introduced by Daya. Before this, we have been communicating with each other for a long time, and today I want to share some personal views with you.

About the future information technology content, including: quantum computing, AI artificial intelligence, block chain. In particular, the connection between the three. I believe that a lot of scholars have been studying these aspects at present, but I am quite excited to have the opportunity to come to this academic conference, because the main content of the conference is the inner connection between the three.

First, share with you a new scientific discovery of the “ancient” science.

Many scientific problems correspond to profound philosophical problems. We exist in a world of unity of opposites. There are positive and negative numbers in this world; there are loans and loans; there are negative and negative people; good people have bad people; angels have demons. There is also a corresponding truth in real life.

In 1928, one of the world’s most famous theoretical physicists, Dirac Paul, tried to apply Einstein’s relativity to quantum mechanics. In the process of deriving the mathematical formula, he has encountered a lot of square root calculations. Then back to the square root problem of your high school, the square root of 9 is not only equal to 3, because the square of 3 is equal to 9, and the square of -3 is equal to 9. So when you start to square a positive number, you get a positive number and a negative number.

Dirac was quite insoluble in the negative number of the founder, which had a profound influence on Dirac. He predicted that all the material in the world had two forms of positive matter and antimatter. Now in Westminster Abbey, you can still find the well preserved Dirac formula.

The “Paul Dirac Medal” obtained in 2012 was the most affirmative for my research work. As I have just said, when you open a number to a number, you can get a positive and negative two numbers, and the negative number is a big rule in the universe. That is to say, as long as it is the matter in the universe, there must be antiparticle. It now seems to be a perfectly normal equation. But in the 1928, that era without antimatter, it was not easy to put forward such a theory. For example, people at the time thought that the antiparticle of the electron was an equal but negative particle — proton, although the charge of the proton was opposite to the electron, but it was more than 2000 times the mass of the electron, so no one was at all believed at that time.

But guess what he said? Lao Tzu’s formula is so ahead of you, your help to learn the slag or to the first class. People do go up. Dirac was very lucky, five years later, when observing cosmic radiation (it was difficult to observe the earth, but in the universe), scientists captured antiparticles and named positron and electron in the same quality but charged the opposite.

I think this is a great hypothesis in human history. The early hypothesis is very good, and the later results are also very powerful. Nowadays, human beings have applied anti particles to medical treatment, such as the famous medical imaging technology, positron tomography, which has been studied by the theory of antiparticle. Hollywood also has a lot of similar film themes, such as the sequel to the Da Vinci code, as well known by the angels and Demons (the original author Dabrown, Tom Hanks), the epic battle of an angel and the devil for antimatter.

Antimatter is seen everywhere in the universe, and if you have antimatter or antiparticle, it’s hard to think of you. But all this is based on a good inference, but the curiosity of mankind has never stopped. After Dirac’s great hypothesis, another great physicist came to our view: ETO Marian Anna. He raised a question: is there any substance that does not contain antiparticle, or does it have a substance itself, itself, antimatter. He puts forward the hypothesis and writes the equation. But this time he was not so lucky, no one believed him, no one had seen the formula, and he was more disappointed. Since then, this problem has become an unsolved mystery in the basic science.

We have a list of problems that need to be solved urgently. What is the God particle and the wave color on this list? Fortunately, there was an answer to the above questions in a laboratory in Geneva in 2012. Again, for example, the gravitational wave, its author, Einstein, is not so lucky as Dirac. 100 years later, it was found that the gravitational wave was a substance, and the gravitational wave was confirmed only two years ago. That’s the magic list of antimatter particles in it. And the matter of antimatter, such as its own problem, is the Mariana Fermi problem, which has been at the top of the list.

Perhaps the search for Mariana Fermi particles is the most interesting problem for scientists. It’s like what I just said. Mariana was very disappointed because no one believed him.

Anna is an Italian. But he went on to the ferry from Napier to Palermo and disappeared. This is an unsolved mystery of the scientific community. It’s been 80 years since he disappeared to this year, but a “good news”: not only did he disappear, but the literature he wrote had not been found. This is the focus of my day. He was writing a formula for derivation, but he did not tell anyone where the formula was placed. That’s why we spent 80 years looking for it, but where did our team at Standford put forward and find out how to find these particles?

Three papers were written by our group from 2010 to 2015. The first is to accurately predict the position of a particle. To our surprise, the particle is not in a large particle accelerator, but in one of our common semiconductor devices. I have already introduced a substance called topological insulators ten years ago by our team, which put some magnetic dopants in the insulators. The material of the topological insulator is some common insulators, and the magnetic dopant is the substance of chromium. So on the surface of the two objects, a superconductor is formed. We think that under this condition, Mariana Fermi particles can be captured but it is not enough. We not only find the location of the particles, but also need to find the amount of physics that needs to be measured, but the common sense helps us. In general, the particle has two sides, like the two sides of the coin, with the front, the opposite, the positive particle and the antiparticle. But Mariana Fermi particles have only one kind of particle: only positive particles, no negative particles. He is a half particle in the legend, so the concept of this half particle will be very important in my next quantum computing speech.

Well, in any case, the Mariana Fermi particle is a half particle of the ordinary particle and there is a certain conductivity or conductivity for the ordinary particle. In general, the conductivity can be polarised by quantum energy, showing that the energy levels of the ordinary particles at the zero order level two level are discrete integers. So, if Mariana Fermi particle is half particle then its energy level characteristic should be the 1/2 level 3/2 level of the ordinary grade. So in a semi – particle system, you can measure the conductivity, but it must be measured at a half – level position.

In collaboration with colleagues at the University of California, California, Losangeles last year, they turned the theoretical model we proposed to reality. On the basis of theoretical study, the energy of Mariana’s particle is measured. In the middle of the one or two integer energy level there will be a half energy level like 1/2. The 1/2 energy level is the key evidence of the Mariana Fermi particle as a half particle, the integral energy level of the ordinary particle. The Mariana Fermi semi – step energy level was published last year by science magazine, and we have a good celebration.

At that moment of excitement, I thought of the movies I had seen – Angels and demons. I thought we had only found an angel in heaven and did not find the devil, so we called the particle an angel particle. So what does it use to say so much? Traditional computers are quite powerful, but what they are good at is not comprehensive. I gave two extra large numbers to add the computer, and he got it second. For example, Gu Geyun, he’s a super fast. But if you give the computer a number, let her produce two related numbers. For example, 15 equals 5 x 3, but 11 is not. You might say 11 equals 1 x 11, but it doesn’t make sense. If a super super large number is given, it breaks down into two related numbers, like decomposition 15, or like 11 can’t be decomposed, the traditional computer is afraid to be surrounded. The only way to do it is to carry on a massive search, and search from 2 to forever.

So what is the biggest problem of computer operation? The essence of a calculation like Gu Geyun is the process of finding the best solution. When the machine searches for the best solution, it searches for all the possibilities, including finding the shortest path. This process takes a lot of time to optimize the search, which takes a lot of time. That’s why computers still need to constantly update their iterations.

Let’s go into the quantum world and see what is the secret of the quantum world. If there is a pair of double seams on the screen, I shoot two slit bullets with an ordinary pistol. Either through the left or through the gap on the right. At the rear of the screen you will see two bullet marks, each one or so. But if you try to shoot two basic particles by double slit, the situation is different. What you see behind is not about two bullet marks, but a complex pattern of interference. When the two particles cross the double seam at the same time, the interference pattern produces, and he crosses the left and right two cracks at the same time. If not at the same time, interference can not be produced. So the quantum world is a parallel world: two particles go through two gaps at the same time.

A lot of people are thinking about the problem. The ability of traditional computers to solve complex problems may be able to use a quantum computer to search all the possible answers at the same time. That is to say, in theory, the computer can search all the possible answers at the same time, and then calculate the answer immediately. It can improve the computing power of the computer so quickly. Think about it.

I’m going to share artificial intelligence with you, and artificial intelligence is a basic concept, which has been proposed in 60s. The reason why artificial intelligence can be developed by leaps and bounds today is mainly the summary of the three big trends. According to the iteration of Moore’s law, every 18 months can double. If quantum computing, not only the 18 months of massage, but a complete change to a qualitative change, our computing power is increasing, almost the same as the past 40 years. In addition, the Internet and the Internet of things produce a large number of large data, large data is to help the machine can really learn, and a good algorithm, and then the powerful computer without data can not reach the best artificial intelligence, in addition to the discovery of intelligent algorithms, and there is a sudden change of rapid progress.

But the whole artificial intelligence, although we see it is changing by leaps and bounds, but I think it is still in a very early stage, its future prospects are still very broad. Why do you say that? Do a simple analogy, like we once saw birds flying, and people want to fly very much, but early learning is just a simple bionics, and we have wings on our own arms. But this is a simple bionics, but the real realm of flight is that we understand that the first principle of flight is aerodynamics, that the mathematical and mathematical equations can be designed for the best flight, and that the aircraft now flies high and fast and good, but it is not like a bird, which is a very core point. Maybe artificial intelligence is now a simple imitation of a person’s neuron, but we should think more, in which there is an opportunity for a major breakthrough in basic science, that is, we really understand the basic principles of intelligence and intelligence, the basic principles of mathematics, so that artificial intelligence can really make a sudden change.

In addition, what kind of criteria are often used to truly measure the standard of human intelligence to meet people’s standards? You may have heard of Turing test, the Turing test is talking about people talking to the machine, but we don’t know whether it’s a man or a machine on the back. In the course of the whole conversation, if you don’t feel at all in a day’s time, that is to say that the robot seems to have reached the level of human being. I don’t agree, although Turing is a great computer scientist, I don’t agree with the decision. Human beings are also the process of evolution. A lot of human emotions are not rational emotions. It may not be so easy to make a rational machine learning a irrational man’s brain. For example, if you may deliberately irritate the robot, it may not be able to manage you.

So I want to put forward a new criterion. What does the machine say, which day really transcends human intelligence? The greatest point of human being is that we can do scientific discoveries. The best criterion is which day the robot can really do scientific discoveries. People know better the scientific discovery, and that day the machine was over human.

I recently wrote an article in artificial intelligence that will be published in the Journal of the Academy of Sciences in the United States. The title is “Atom2Vec”, one of the greatest scientific discoveries of human beings, the theory of relativity and quantum mechanics. The greatest discovery in chemistry is the discovery of the periodic table of elements. Today’s machines assume that we don’t know the element periodic table at all. Today’s machine can automatically discover the periodic table of the element without any guidance. What we have entered is the name of all the existing scientific elements, and the names of these compounds are entered into the algorithm. As a result, the machine naturally finds the periodic table of elements, which can make the greatest scientific discovery that humans think. Then our program can help us find new drugs, and we can find new materials by machine learning.

Next, I will share with you my last topic block chain, artificial intelligence is developing rapidly, but the most lack of artificial intelligence is data, precisely today data is in a state of complete central monopoly, and can not help machines to learn reasonably. It’s heard that a lot of Facebook personal data were stolen a week ago, at least without being allowed. In today’s world, individuals will produce a lot of data, personal genetic data, medical data, educational data, behavioral data, and so on. But a lot of these data are in the center of the center, and there is no real centralization. But the production of a block chain is the ability to produce a data market. So my ideal world, the future each has its own data, this is a completely centralization of storage, so that hackers can not black everyone’s data. Then some encrypted algorithms can be used to protect the privacy of the individual and make good calculations on the block chain.

So I describe the whole idea of the future block chain in one sentence, called “In Math we trust”, and our beliefs are built in mathematics. It should be remembered that I have seen a lot of people taking pictures, and one day it will certainly be the most fundamental of the whole block chain and the whole IT field, which is both the most basic mathematics, and can lead to the protection of personal privacy in the data market and the ability to make reasonable statistical calculations. For example, there is a very magical computing method called zero knowledge proof, which can prove to you that my data is very valuable, but it does not tell you where the real privacy is.

The main topic I report today is a core idea that is to make IT really able to develop, both in physics and in mathematics. Shenzhen has done well in its applications, but since the university is not the most leading university in the world, I suggest that the most core investment, this kind of mathematics and physics, is really closely linked to the IT field.

After a block chain, the production of this data market, we can really make society more fair, and the biggest injustice in our society now is that we can easily discriminate against some minority. But what is most needed in the process of machine learning is the data that the minorities have. If the accuracy of today’s machine learning is up to 90%, I want to make 90% to 99%. It needs not the data that has been learned, but the most different data from the previous. The data that is often owned by the minority is the most valuable for machine learning. Once we build on the basis of the block chain and add these wonderful mathematical algorithms, we can really reach the data market, which is the most valuable data in the data market. In this way, we can really turn an ugly duckling into a beautiful swan, because the ugly duckling is not ugly, just different from other people. In this world, the world idea that the block chain and artificial intelligence coexist with each other is the most valuable.

The understanding of the whole block chain is not the most fundamental principle of the first nature. In terms of the basic principles of physics, the consensus is that everyone agrees with the same book, which is in physics, for example, the magnets are originally disorderly, but in the ferromagnetic state they are the same direction. So the consensus is in the natural world, and in today’s human world. But this phenomenon is called the phenomenon of entropy reduction, reaching consensus, and everyone is going in one direction, and the entropy of this state is far smaller than the chaotic entropy. It is very difficult to achieve this consensus, because entropy is always increasing, and it is very difficult for you to reduce it today. The ability to reach a consensus system on a block chain is an algorithm that requires energy consumption. People may not understand at first why it sounds unreasonable, and why some accounts take energy. From the second theorem of physics, this is a very reasonable thing, because the consensus itself is entropy reduction, but the entropy of the whole world must be increased, so we must exclude some other entropy at the same time. The non centralization of the mechanism is very similar to the magnets in the natural world from the chaotic state to the orderly ferromagnetic state, and the cost is also the inevitable trend.

I share with you that, in addition to being a professor of Stanford University, I am also a founder of the capital of Dan Hua. Our main core concept is to link today’s most advanced technology and investment, to understand the world of today with the way of thinking of the first principles.

The other thing I want to say is that I am from the academic world. We need to do two big bridges in the field of artificial intelligence. One is to make close connections between the academia and the industry. We have the best physics, the best math and algorithm discovery and invention in the academic world. In January 8th of this year, it was a great honor for me to receive the People’s Republic of China international scientific and Technological Cooperation Award from the general secretary of the people’s Congress at the general assembly hall. Our whole world science is the most endless, the least national boundaries. Science can truly bring mankind beyond the borders. Today, the artificial intelligence and quantum computing we are going to solve are all human problems. So we must not put our eyes on our own part, but look at the whole world and the whole world. In this process, China is also a very big opportunity. All of us want to answer the question. In addition to doing well the application of science and technology, China can have real original science and technology in China. These are the most basic physical and basic mathematical principles that I have introduced to you today, and we can do it well, and these principles sound more abstract, such as entropy increasing principle, positive and negative electrons. But at the most basic level, this is the wonder of our world today, and it really can provide the most basic prospects for the development of science and technology for the entire IT industry. Thank you.

The following is the question and answer link:

“So you mentioned the consensus, and the work proof system achieved consensus, through decentralization and by increasing unity.”


“So how does it get in the space unified system?”

“Well, in fact, I think there will always be some trade-offs at the end. I think the future of the block chain and the cipher currency will be similar to our present world. Now the world will have M0, M1, M2, different levels, so I think the most basic layer. Global currency should be entirely on the basis of work proof. Because then the unity you dumped is completely transparent. It’s not just because it has to be there, but it must be completely transparent. I think the most basic and fundamental layer, state proof will not be useful, because there are too many collusion possibilities. You will lose some of the links, but also get some unlinked things, there will be bribery, so I think the block chain is the most exciting thing in the world.

It is in the most fundamental layer that there is something completely objective, and that it is only connected to the real world, that is, energy. If the discord is connected, it will be filled with human irrationality. But I can imagine that at the higher level they will be useful, but at the most basic level. The state of M1 or M0 needs a state of complete stability

So I still think that work proof, or another way, is called space time proof, space proof. On the basis of capacity, I think it is also a quantifiable physical resource, and I think it is the least human thing to be included in it, but maybe it is the most included.

I usually think that quantum computing is a useful search algorithm for artificial intelligence, so one of the most interesting ways of artificial intelligence is that it is a useful search algorithm. So I’m not saying that these three trends have to work together forever. In fact, they can make progress through competition.

So on the one hand, quantum computing and block chains are competing with each other, because the chridge coding algorithm may be broken by Compton. But on the other hand, I see that Compton can help AI do the most efficient search, which is what AI needs to do. So this relationship is like a symbiotic relationship in our ecological environment.

We can’t use the human idea to assume that they will do the same thing forever. I think that in the process of competition, they will become stronger. “

“You mention global currency or M0 or 1, I am curious, I know you are a theoretical physicist, but in the process of execution, or take iPhone for example: my iPhone7, encountered iPhone6, encountered iPhone5, but there is a metal layer consensus need to be reached, and I actually agree with this distributed system.”


“There are a lot of debris pools in cryptography, and the proof of liquidity is not cited. So how do you build bridges on the gap we are now in? “

“For example, the relationship between the bitcoin block chain and the lightning network is very suitable for the framework of the M1 M2, so basically the block chain is completely objective under the work proof, so it tries to achieve the most universal consensus among the parties. Parties are completely unaware of each other, but still need to negotiate and communicate, but when you really think about business, maybe two of us have been happy as partners in the last ten years, so why do we still treat each other as a stranger? So what we can do is get into each other’s state channel and put our carat on the block chain, but we still do a very, very fast deal, but we still pick the load once a month. I think it’s like the relationship between M0, M1, M2, and the relationship between lightning and bitcoin, like the relationship between M0 and M1.

So when you go on each layer, the stability will decrease, the efficiency will increase, but weighing the history from us, we have a history of trust, so if you have business partners, they are familiar with each other, and they do not need to use the most popular solid layer. They can build higher levels and sacrifice some universality, but in exchange for efficiency. “

“I have a question about angel particles.”


“Angel particles are not positive?”

“Yes, it’s half the bit.”

“It sounds like an authenticate in the actual field, when faced with your own problem…”

“No, in a more accurate analogy, it is like a string of complex numbers that can be expressed as two real numbers, so complex numbers are like particles that are complex as antiprime particles. If you have a real number, the complex is the same as it is, so the angel particles are more like a real number. “

“I see. We think this is the contrast between yin and Yang. “

“Yin and Yang, right.”

“What can be an intermediate element, what is an angel…?”

“I think the analogy is just that there is a coming qubit here, but the actual calculation, before you do the actual calculation, you are splitting them, but by splitting them, they are already nonlocal, they are entangled, but the classic interferon is not chaotic, so it can’t be removed from the classic interferon. That’s why topological quantum computers are so stable.


“Well, so combine the themes of your speech: if you can manage the energy of a quantum computer, then if we can manage our data safely, we can share data in a privacy – encrypted way.”


“I want to know what you think about the future of Google. Because it’s like a real threat. If anyone can black into a quantum computer that does an efficient search operation. “


“Then they can control any person’s data.”

“I think the only way to do it is not to resist change, but to embrace change.”


“So how do you see Google mastering this technology in the future world?”

“In fact, I have an answer: so in fact, we can do the following. For example, my personal data, I want to store it in a safe way, but still can do some operations. We know that the Google cloud is competing with the Amazon cloud, so what we can do is to store completely random numbers in the Amazon cloud, but on the Google cloud, I store my personal information, plus the random information I store in the Amazon cloud, so if I can’t predict that these two very competitive entities will not have collusion, they can’t exchange information, then you can use multiple security computing systems to do the calculation. A computing result is obtained without leaking any details, so in this world, the concentrated entity is still useful. But in order to make it feasible, you have to assume that both sides are competing, not colluding.

“Hello, I think the use of terminology is very interesting. Because it seems to be a mysterious thing, but it is also valuable. In thermodynamics, you can have a logarithm in the classical thermodynamics, and then you have a unified information theory, and then you use energy to do the analogy, which reminds me of some free energy. “

“Yes, it is. I think the block chain world does get some free energy from it, so you basically get something, but no matter what amount of energy you get, the amount of useful energy you use, the energy you use minus the unity you have to waste. So today you can still see a lot of whites who claim to create miracles, and these whites remind me of the perpetual motive for the eighteenth Century.

“I want to know that you can further speculate on the analogy, and you need a temperature to make it work.”

“Right. In fact, the temperature appears to be very natural, and when you have a saving amount, such as energy conservation, the concept of temperature naturally occurs. Because whenever you have a random but frugal system, this is the most common, known as the Boltzmann distribution. So the introduction of temperature is very natural, but why I am so excited about this is because it is the first time I see the convergence of social science and natural science, and it provides a dependency on the social science world, so I think M0, M1, M2, as the foundation anchor.

Now into the natural science, we can accurately see the unity, it is wasted. So we can know why it reaches the consensus, and then you build more human things on the basis of it. But at the most basic level, it is now a common understanding between the social and natural sciences, which fundamentally reduces energy, unity and information.

“Thank you so much for taking so much time. I think in your speech, you’re saying: you’re going to see the first layer of a block chain, and then more layers are built on this basis, so you think all kinds of projects and companies are trying to build their own block chain, which is related to your speech? “

“I think you offer something unique. So bitcoin, block chains and theorems are very different, so as a base of trust, you don’t want the universal Turing machine. Because maybe it will be black, but you have to do more on the basis of it, and if the theorem looks more natural, the evolution of the block chain world will be like the evolution of the biological species. You see branches, different species, and if they have been forked long enough, they may become different species, but there is always the most basic thing. All organisms are made up of cells, so this basic contract will not change. But for the organization, it is possible for different organisms to be changed by different tissues of different cells. “

“OK. Thank you for your explanation. “

“My question is: when do you think quantum computing can be applied actually? After your discovery and research, and then when it is actually applied, do you think it will only be mastered by a particular big company? “

“Good. I think the study of quantum computing, the most idealized words should be open to use. I think, let me make a statement: I know that a lot of companies are trying, but the basic meaning of the company is that they want to protect the interests of shareholders, they want to protect their secrets, but for this kind of huge energy, and what is not known to the human impact, I think it is best in public research, and this is what I am doing.

So the way I deal with the quantum computer, there are very, very many companies trying to join my quantum computer research, but I’m against that. “

“What do you predict for the use of quantum computers?”

“Does it contain or do not include my invention?” I think if you try this way, it will take a lot of time, can you imagine? You need seventy bits to serve a useful bit, I don’t think it’s worth it. But in this way, it’s worth it. “

“Well, I think we’re about to end, and I ask the last question: about your angel fermion, does it change any other requirement of quantum computing, such as absolute zero?”

“No no no, no, it still runs and runs at a low temperature most of the time, but it’s a pity that our method can run at room temperature, if we find superconductors that can run at room temperature. If there is really substantial improvement, we can drop it to low temperature. “

“Thank you very much, Professor Zhang.”

Leave a Reply

Your email address will not be published. Required fields are marked *