Keynote: Professor Shoucheng Zhang on Quantum Physics, AI, and Blockchain | SVIEF
9:57PM Sep 29, 2018
So our next speaker is a world leading scholar in quantum computing, AI and blockchain, and also a good friend of SVIEF, a man of many roles. Not only is he a well respected physics professor at Stanford University, he's a member of the US National Academy of Science, the American Academy of Arts and Sciences, and also a foreign member of the Chinese Academy of Sciences. Now he discovered a new state of matter called topological insulator. Now for this ground breaking work, he's received numerous international awards. Well, just to name a few, Buckley Prize, Physics Frontiers Prize and the Benjamin Franklin medal. He's also the founding chairman of DHVC. Now, ladies and gentlemen. Let's give it up for Professor Shoucheng Zhang.
Welcome Professor Zhang.
Thank you very much for the introduction. Every year, I look forward to these great event. I have been coming here for the last three years and have been chairman of the SVIEF for quite a few number of times. So today I like to give my lecture in English and to the slides will be in Chinese. So hopefully, nobody need any translation.
So I'd like to tell you today presumably all of you come to here congregate here in Silicon Valley and you'll wonder what's the next generation of technology which will shape our future, let's say, for the next 20, 30 or even 50 years.
So I think we're resting on three pillars of information technology, namely quantum computing, AI and blockchain. So I like to use my brief 15 minutes to tell you a little about each one of them.
So first I like to tell you about scientific discovery made my by made by my group at Stanford University about Angel particle cancer leads. So we know anti also has deep. It's a very interesting and curious things scientific research, but also has profound impact on quantum computing.
So looks like our world is made out of opposites. We have positive number, and negative number, credits, debts, yin and yan, and angels and demons.
So, in the world of fundamental physics, this philosophy of the opposites also have a direct implication. And namely this great physicist Paul Dirac predicted that for every particle in the universe, there exists an anti particle.
So this was his remarkable prediction maybe in the effort of unifying Einstein's special theory of relativity with quantum mechanics. In the process of doing so he predicted anti particle and only five years later, the anti particle of the electron, namely the positron was discovered in cosmic ray radiation. So this has been one of the greatest prediction in human history ever.
So it also has caught the imagination of Hollywood's. So many of you have watched the movie or read a book of the Da Vinci Code and there's also a sequel to it. And this is called Angels and Demons. It basically depicts the epic struggle between angels and demons, and comparing that with the annihilation of particle with anti particle, of the electron with positron, and the tremendous amount of energy that's released where matters are destroyed and convert them completely into energy. And it's due to the particle and anti particle annihilation process.
But human curiosity don't quite stop there. So it looks like it's a universal philosophy, both in the east and the west, the philosophy of the opposite. It has also the analog in the world of physics for every particle they exist a anti particle. Between 1937 more than 81 years ago, a great Italian physicist conjectured that could exist a particle which is its own antiparticle, a particle which does not have a anti particle. So it was a pure theoretical conjecture and he didn't prove provided any hint where to find it.
But even more mysteriously exactly 80 years ago or one year after he made his prediction. He disappeared. He took a ferry from Naples, the Italian port city of Naples to Palermo his hometown of Sicily and he never came out of the ferry. So this great physicist disappeared and his particle was elusive.
So if there's a most wanted list of all the dreams of fundamental science about missing particles we look for, this is certainly very high on the list. For example, in 19 2012 we discovered a long conjectured particle called the god particle or the Higgs particle and it was finally found. And only one year later it was awarded the Nobel Prize in Physics. But then there are other things we look for such as magnetic monopole, dark matter, but this Majorana fermion is also a very, very high on everybody's list.
So in the last many at about 13 years ago, my group started looking for this particle and we predicted to very, very precisely where in which experimental condition we can find this angel particle. And this is in the system of what is called a topological insulator doped with magnetic dopants and on top of it you apply a superconductor.
So now many of you may wonder what does it exactly mean to discover a particle? Because even though we know the world is made out of atoms. No one has really seen an atom with the naked eye. So you kind of need to find the experimental signature to find what is the signature for this new missing particle. So there's the idea that for every particle usually there's a anti particle. But for this Majorana particle it has no antiparticle. So it's in very some sense that this is one half of a regular particle.
So it turns out the regular particle has some quantization phenomena. They can only take value between 01234 and so on. So, but if this Majorana particle were to exist, which is a half of a regular particle, then we should to predict some plateau at half integer value, such as one half, three half and so on.
So last year, there was a great excitement in the scientific community, my group at Stanford University work with a group at UCLA, UC Davis, and UC Irvine, and they perform the experiment and indeed they find there's a signature at one half of the usual value and they exist a new plateau and that plateau is the indication of the discovery of this new particle.
So many of you may wonder what is the pleasure of being a scientist. And most of the time we have to work very hard. Professors usually are not paid very much. So what exactly is the pleasure of being a scientist? Of course it is the joy when you become the first one in the world to discover something. But just like giving birth to a child, it's like a discovery that no one else has made before, you have the right to name your child. So in the case of physics or science, whenever we make a discovery we'll also earn the right to give a name to the discovery.
So I remembered my favorite novel novel which is Angels and Demons, and to the epic struggle between angels and demons. And I suddenly realized that, as if we have discovered paradise with only angels and no demons, only particle no antiparticle. So therefore I give it a name called Angel particle. It was explosive interest last year about this discovery of Angel particle.
Now you may wonder what is useful. So in in computers we have it's very easy to multiply two very big numbers to get a number, but sometimes you also like to ask the question whether a number is more like the number 11 or more like the number 15.
So what's the difference 15 is equal to three times five. But 11 cannot be written as a product of two smaller numbers. So this is the problem. So if somebody gives you a very, very large number, it's very hard to tell whether it is more like 11, or more like 15, whether it's divisible by another number or not. And the best way to do it in the classical computer is trying to divide it by all possible small integers. And then you can find out. But that takes a very, very, very long time.
So then enter the world of quantum physics. In quantum physics, a particle actually can go through two slits at the same time. If I have a classical particle, it only goes to the right of goes to the left. But the quantum particles can go through both slits at the same time. So the quantum world is intrinsically parallel. And therefore, if you use quantum mechanics, you can do this problem very, very quickly. Because you can enumerate all possibilities at once in parallel. And that help you to solve this problem very, very fast.
So everybody like to build a quantum computer, Google, IBM, Intel, Microsoft, and so on. But I think their approach is a little bit flawed. And they're using approach like building analog quantum computer, which has a lot of errors and mistakes. But after the discovery of for discovery of Angel particle, we realized that we can build something like a digital quantum computer, that it doesn't make the error as much as the analog quantum computer.
And so this is a very exciting development. And one of the greatest pursuit of science and technology today, especially when you come here to Silicon Valley is the pursuit of building a quantum computer and told you a little bit about my story.
So now, let me tell you a little bit about artificial intelligence. So it's mostly because of the magic conference of three major trends.
One is Moore's law, which says, all computational capability doubles every 18 months. And the other is the availability of large data set. And the other is to computational algorithm. So on the left here, or on your right, you already see the magic of artificial intelligence in transcribing my speech by our very proud investment in otter.ai.
So now, there's a standard test called the Turing test, which is trying to figure out when does the AI reach the level of human intelligence by making a conversation to the AI or to the human. And if you cannot tell the difference, Alan Turing would proclaim, then the humans then the AI have reached human level.
So I think this test is little bit subjective. And I have formulated another test, which is totally objective. And that's to ask whether AI can make a scientific discovery, a great scientific discovery before the humans do.
And actually my group experimented with that, for example, one of the biggest discovery in human all of human history is the discovery of the periodic table. But very recently, or group has introduced a new computational algorithm called atom to vec, which is modeled after the natural language processing AI algorithm called word to vec. So basically, by feeding the name of all chemical compounds to the AI, AI can actually figure out and map each element or atom onto a vector. And then when you project this vector onto two dimension, we discover exactly the periodic table. So basically, AI is capable of making a great scientific discovery. And so we just showed one example that AI can discover the entire periodic table in a matter of two minutes, whether whereas it takes human minds about two centuries to discover the periodic table. So we're generating this algorithm and to maybe use it to discover new or interesting drugs.
So now, let me tell you a little bit about my thoughts about blockchain. And I'm very fortunate to come right after Tim Draper who shared with you his tremendous enthusiasm about blockchain. So I think the history of the world and the history of the networking world can all be characterized by eight Chinese characters, fen jiu bi he he jiu bi fen, from centralization to the decentralization.
Very early on, for example, we have AT&T which monopolized and centralized all networking resources. But when the TCP IP protocol gets invented, immediately, it became decentralized. And the monopoly of AT&T gets removed or destroyed. But then the content get very much spread all around the internet, and very hard for anyone to find. And then a centralization come up again. And so there's entities like Google and Facebook, which we organize this information and making them more easier for you to find. It's a reorganization of the information.
But I believe that with the introduction of the blockchain, finally, we're seeing another decentralization wave. So all of you, especially all of you, entrepreneurs here must feel very, very lucky. Because a few years ago, you may say, Well, I'm so unfortunate that I missed an opportunity of making a startup like Google and Facebook or like Baidu, Ali, or Tencent. But I'm telling you that your opportunity today is much greater, because just like Tim, I believe the blockchain opportunity is maybe 10 times greater than that of the internet, its impact will be really, really profound.
So maybe, let me tell you that the fundamental idea is that you can have a decentralization mechanism. So by reaching consensus. So if we have a vast population, it's usually very difficult to reach a consensus about something, for example, about when and to when certain transaction has occurred. And in order to reach this consensus, the blockchain world uses an algorithm called proof of work. And this is very natural, because reaching consensus, in some sense, reduces entropy. But the world always tends to increase entropy. But by proof of work, you're actually burning some extra entropy in realizing consensus state of less entropy.
So I find this is a brilliant, interesting new design that's purely based on the principle of mathematics and physics that we can reach, find a new mechanism of each consensus.
So I think very big opportunity today is AI plus blockchain. What does AI need the most AI needs data. But so far data marketplace hasn't really emerged, because it's quite unclear who owns the data? And if you own the data, how do you actually monetize on that. So I think blockchains ideal in keeping a data marketplace so that individuals can own the data. And then whenever I as an individual contribute to the machine learning by making my data available, then I will get incentives through the blockchain ledger.
So all this is enabled by a very interesting new set of technology. And I will characterize the future of the world was one slogan only, and namely, in math we trust. We can build trust previously, it's very difficult to build trust. But finally, we can use math to build trust.
So as they're saying that great minds think alike. So I am very fortunate to come right after Tim Draper. I think two of us have never synced about our talks. But actually, I just started a company called novo vivo. And the idea is exactly what Tim tries to explain to you, that each individual one of us should own all of our biomedical data, my genomics data, IoT data, and my electronic records. So all of these will be belong to me. And it can be stored somewhere on the cloud, but with full encryption. So that's only I hold the private key. But the magic of this math, of this set of mathematics is that you can compute on encrypted data.
So for example, a pharmaceutical company can compute on my encrypted data with a large population of which larger cohort and learn how to develop a better drug. But in the process, they will pay for the data contributors, as kept by a blockchain. So why actually making such a company with a lot of exciting investors and so we're just trying to make these into a reality.
So let me conclude. And maybe that blockchain is one of the most interesting development of our time. I think it's 10 times bigger than the internet and all throughout history and to you will see that these centralization and decentralization waves are what really propels the next generation of technology and there's tremendous opportunity ahead. Thank you very much.