Editor's note: This article is the report of the president of the HKUST and the president of the institute, Hu Yu, at the Sina C+ speech. The editor of Lei Feng Wang (searching for "Lei Feng Net" public concern) did not change his original intention.
The topic of my speech today that I would like to share with everyone is " From a candid to a cognitive revolution where one can understand and think and a speech as the entrance ." In the past two years, it should be said that everyone suddenly became very hot for artificial intelligence. It should be said that many people are excited. And a lot of problems have arisen. We all think that artificial intelligence is just like the host just said. In March, it defeated the world Go master Shishi Shi. Many people use this to infer that artificial intelligence is to subvert humanity and even to extinguish mankind. But is it really true? Let's take a look together.
| What is artificial intelligence?When it comes to artificial intelligence, I think it's important to figure out the first thing. What is artificial intelligence? How far is artificial intelligence from us? Let's take a look at the first topic. Everyone is talking about where AI is where he was born and what kind of development it has undergone. We all know that the topic of artificial intelligence must be traced back to the birth of the world’s first computer, Eniak , in 1946. After the creation of Eniac, many computer scientists will be able to replace humans in the future. There are so many associations for things. The most famous one is Turing. Turing invented a very important bomb during World War II. This bomb could break the German cryptosystem at that time. After the Second World War, after the invention of Aniak , Turing conducted many discussions in the field of artificial intelligence around 1950 and proposed the famous Turing test.
But the word "artificial intelligence" was put forward at the 1956 Dartmouth meeting. This year is the 60th anniversary of the term "artificial intelligence" being proposed. We can see that, whether it is the artificial intelligence conference in the world or the artificial intelligence conference in China, it may be that the term artificial intelligence conference will not be unheard of. Those who brought artificial intelligence at Dartmouth, including Minsky, Simon, McCaset, etc., these young artificial scholars after they put forward the concept, they were born four Turing Award winners. As you can see, the time has come to pass as the last participant, but we can already see the development of artificial intelligence.
However, the development of artificial intelligence is not always easy. Like a technology, when a new technology emerges in human society, everyone will have very high expectations. When artificial intelligence was just put forward, everyone believed that artificial intelligence could solve this problem by using rule-based reasoning . Because of the computer, storage, and various reasons at the time, it was the era of mainframes. The president of IBM once said that the world only needs five computers and it turns out that he was wrong. However, after artificial intelligence developed in the 1950s and 1960s, it was far from everyone's expectations, so artificial intelligence entered a period of silence.
With the development of personal computers in the 1980s, artificial intelligence once again attracted everyone's attention. Taking the fifth-generation computer in Japan as an important performance, in China's 863 plan, in fact, we also designed a 306 theme, which is the theme of a smart computer. China’s leading world supercomputers, including HKUST’s own flight, benefit from the investment and innovation of the National 863 Program under the theme of smart computers.
But in the 1990s, artificial intelligence once again fell into the doldrums, because many people found that the Internet can actually help us solve many problems. Our information technology can solve most of the problems in our lives, and we do not need artificial intelligence. And artificial intelligence can't do it either.
Entering the 21st century, especially with the development of the mobile Internet and the Internet, we once again discovered that artificial intelligence is actually possible to do. Wait a minute and I will explain to you why you can do it.
| Does artificial intelligence threaten humans?This is a prediction made by Gatner, including Deloitte's analysis. Many people are very interested. Now a very important topic is whether artificial intelligence will threaten humanity and will not grab our jobs . As you can see, the report on the far left is a report from Gartner and it is said that artificial intelligence is increasingly participating in our lives. The Deloitte report in the middle has analyzed the current human work, and how much has been replaced by the artificial intelligence technology that is now available. The good news is that only about 20% of CEO jobs can be replaced. There is also some good news that the proportion of our gardeners and cleaners being replaced is even lower, only 10%. But there is also a bad news. For all of you here today, you can see that the robot can directly translate the voice into text. This is a good thing for reporters. But at the same time, it will also start writing. As you can see, the recent United States, including Tencent's artificial intelligence technologies, have already been demonstrated. If robots are reported, the declarative manuscript may be more accurate than humans. .
In fact, we can also see that the development of artificial intelligence is not a linear one. This is a report last year or this year. I think there is indeed such a possibility. Everyone knows that our universe today is actually more than 100 billion. There was a classic big explosion a year ago. But has the development of artificial intelligence been faced with the same situation after 60 years of ups and downs? Many people believe that including Ray Corswell, many people call him a liar, saying that artificial intelligence will have an explosive development in 2045. It turns out that in the digital world, it is true that artificial intelligence is likely to change at a basic point. But we also know a lot of real things, such as our energy, materials, and physical world are not so big. Therefore, I think this may be a very subversive issue. We will make surprising changes in the next 20 years, but will it subvert the human race? I think one of the topics in today’s C+ summit is very interesting. It is consciousness. I will wait for it later.
In the past few years, everyone has accepted the Internet and mobile Internet as an infrastructure, just as we once entered water and electricity into our lives. Now everyone can see that in the next 10 years, artificial intelligence will also be like technology services into our lives. Everyone will be able to leave. The reason why its artificial intelligence receives so much attention is not only because of this technology, but also affects our lives. It affects not only each of us but also our country. As far as I know, U.S. Defense Secretary Carter recently went to Silicon Valley because the U.S. wants to continue to maintain hegemonic status in the future. It is very important that he wants to take the lead in military production of unmanned combat vehicles and unmanned warships. Troops. Everyone knows that artificial intelligence talents are very expensive, so Carter wants to go to Silicon Valley to find someone. That is a contestable position in the country.
| Artificial intelligence Overcoming Difficulties in what?When it comes to artificial intelligence, many people have all sorts of ideas, but artificial intelligence is the most important. Now, the most important thing is to break through, and where is the most difficult place? To answer this question, in fact, we may have to look at humans first. We all know that humans have 300 to 4 million years on Earth. In March of last year, in 2015, for the first time in this forum, I explained to you the difference between human intelligence and artificial intelligence.
The author of this book, Jacques Attali, should now be known to have become famous. Everyone knows, but no one brought him together with technology when he first came out. This is a book written by Jacques Attali, a very young historian in Italy, called A Brief History of Humanity. He proposed some very important modern modern archaeological developments. Because of the development of genetic technology, we know all the humans on Earth, including white, yellow, black, and brown races. Our ancestors are all Sage from Africa. Before the Homo sapiens, there were three or four million years of human monks on the earth. In the Paleolithic and Neolithic eras, we had Yuanmou, Peak Cave, Lam Tin, etc. in China, but he was not ours. ancestor. Why did humans 70,000 years ago, we now see that all human ancestors, wise men, have exterminated other monks in the world and acquired the status of the true ruler of the earth.
We can see that in fact, the book must also be very clear, that is, 70,000 to 20,000 years ago, human beings had an important revolution in the agricultural revolution and industrial revolution as cognitive revolution . The cognitive revolution is a very important aspect of human beings' victory over other deaf people with language and intelligence because there are several aspects. One is that they can accurately describe objective events; the second is gossip. I used to think that gossip was a feminine society. Good habits, but I did not expect to be very important, from the original 100 people's team can be changed into a team of 150 to 1000 people. Because of the language, we can communicate what we are dreaming about, or what we think. We can produce fictional concepts. We can produce religion, public and private, and so on, so that we humans become the masters of the earth. But this old man, Jacque Atari, recently wrote a book and has already announced that he is about "A Brief History of the Future."
I once said that because our current artificial intelligence has nothing to do with our consciousness, I do not think that the development of the machine will be able to mutilate humans or replace humans in the future, but Jacques Attali said that even machines will not produce Self-awareness, when its intelligence reaches a certain level, all humans will be lazy to lie there waiting to die. Of course, I very much disagree with this point. I want to see what the "A Brief History of the Future" will be after it is written.
We can see that the media that have learned a lot in the past few years have accepted such a view on the classification of artificial intelligence . This is what I came up with. I referred to it as computational intelligence, perceptual intelligence, and cognitive intelligence . After the invention of the machine, we all know that computing power and storage capacity far exceed humanity. In terms of computational intelligence, machines do far exceed humans. In the "human-machine wars" in 1997, they defeated the super-parallel computer "dark blue" of the world chess champion Kasparov. For example, the Alpha dog saw 30 million chess games, of which more than 29 million were his own and felt the world. We can see that we can hear that we can perceive. The machine is like an animal. The machine is catching up with us quickly because the machine's sensors and artificial intelligence algorithms become more and more clear. In terms of cognition, animals are far from humans, and political development is very limited. Without language, there is no knowledge. Without knowledge, complex reasoning cannot be done. So in terms of cognition, if machines can catch up with us, it is a very important node for machine intelligence to overcome human beings . There is still a big gap between them.
It is precisely because of this that we can see that its pathway is actually divided into two kinds. One is the major artificial intelligence program of the brain sciences of the countries we are participating in. We hope that through the understanding of the human brain, we can help us design. Better algorithm . But we all know that understanding the brain is a very long process. We are still working hard. We and the Department of Neuroscience of the Chinese Academy of Sciences are in contact with many departments. But on the other hand, the artificial intelligence algorithms that we have used in the industry today should actually be called big data artificial intelligence. It can be learned through many of our technologies.
I don't know if you have recently read the description of this aspect. It is about what we can learn in the brain science. Once Wu Jun had a book called "Intelligent Age". There was a reference to it called "Bird Flying School." When we wanted to learn to fly, we first thought of flying like a bird. We put a feather on it to find one. A higher place to jump, I believe we all know. But then humans discovered aerodynamics and we made better equipment than birds. When we study the brain, we do not have to completely copy and clone the brain. We can find the intelligence here . I believe we can find out about the mechanism, mechanism, and structure of brain learning that can help us.
What is the idea of ​​artificial intelligence?But on the other hand, we can actually see what really can be seen in the current industry. We use the deep neural network, big data, and our ripple effects to jointly develop our current visual, auditory, and natural language processing abilities. In this respect, I would like to introduce the ripple effect . The ripple effect and network effect are somewhat similar. It is the thinking of using our Internet thinking to solve artificial intelligence . Because everyone knows that any artificial intelligence comes out when it is not so advanced, but it requires real people to use it, to contribute real data and experience, to make a product into this. When a water drop drops to the surface of the water, it takes time to spread. The more time it spreads, the more people use it. The water drop has the wrong error. The initial error is very large. As more people use it, the technology improves, and the wider the spread, the smaller the error. When it didn't spread to the entire crowd, it was well trained and the water ripples have slowly disappeared. But for those who have just come into contact with it, they will feel that this thing is so good. This is the idea that we are now doing artificial intelligence.
What is the HKUST flight doing?Finally, I would like to talk about what the HKUST is doing in the end. It is precisely because of these foundations of the previous theories and our vision of the future. One of the things that HKUST is doing is that XFX is a brain , and we are going to do voice and language as the entrance to the machine. The cognitive revolution includes perception intelligence. And cognitive functions . This is a picture of our entire XFN super brain. This is to imitate our human perception intelligence. After this, we must rise to the level of human cognition. We have a Think Machine that can carry out your concept. Reasoning and solving . After the solution was completed, there was an implementation. In fact, we have achieved very good results in this technology recently. Let me introduce it to you.
It is because of the fact that HKUST is flying in the deep neural network. Everyone knows that we invited an alumnus of HKUST in the first two days of the China Artificial Intelligence Conference held in Beijing, that is, the National Engineering Laboratory of our Xunko Language and Language Information Processing. Professor Deng, who is a well-known international scholar who worked with Microsoft to make neural networks work together successfully. Three months after their success, it was in December 2009 that we knew that deep neural networks could actually be used for our artificial intelligence calculations. In the past five to six years, HKUST has used neural networks for the first time in speech and language, such as speech synthesis, language recognition, language recognition, natural language processing, and so on.
| How to get results?What kind of results have we achieved now? Everyone knows that our voice is actually in our mind and we can see a voice spectrum in the voice processing. Very cattle personnel, use this spectral map to judge. The results we are now doing can use image processing to process our speech, which is very close to the hearing system in our ears. Not only that, it is more important to solve cognitive problems, including speech translation, knowledge representation, association reasoning, and autonomous learning . We also have a lot of core modules, which are not introduced here one by one, including autonomous semantic representations and logical reasoning.
What kind of results have we achieved now? We have also borrowed a very important attention model from the human brain. The figure on the right is more interesting to everyone. The left side is a map of the lighthouse. The top of the right is the place where the man looks and pays attention while driving. The following picture is alleged to be the view of the woman when driving, and the field of vision is very narrow. Therefore, most female compatriots may have more accidents.
One of our newly released products is AIUI, and we very much hope to have very natural interaction with all smart hardware. But we all know that people who have used voice interactions, although the Infofly voice cloud now provides more than 2.5 billion voice services per day to 180,000 developers, and we also connect to 80 to 900 million devices, but their It feels frankly not good. Using the AIUI, we have no way to demonstrate because of time constraints. We turn the original single interaction into full-intelligence interaction. We integrate speech recognition, speech synthesis, natural language understanding, and the entire dialogue process. Eighty-nine percent of domestic products are now connected to our new AIUI products, bringing us different experiences.
As you can see, after using this technology, the success rate of our dialogue has increased from 40% to over 90%. In addition, two achievements in the perception of cognition by XFN super brain, I will give you a brief introduction. One is about spoken language assessment. Everyone knows that the most painful thing in learning English in the world is that you will find that there are no oral tests in the world. Why? Because there is no way to change the score in the speaking test. Guangdong Province once had an oral test in the college entrance examination. After 770,000 candidates recorded the sounds, they found 2,000 English teachers and teachers and students to listen to. They did not change after two weeks, and the pressure was very great. However, we now use the machine method to automatically understand what is spoken in your spoken language and give you an assessment. If you read wrong, your voice and semantics can be evaluated. Now only China is in the world. Guangxi, Guangdong, and Jiangsu can take the test of spoken English. Not only can we test spoken English, we can also write and write essays. We can evaluate and revise your composition. I am still leading a team of 863 experts and should be said to be the largest team in China. 31 units attended, hundreds of teachers, hundreds of professors participating in the Chinese college entrance examination robot, that is, we have to do a robot reference college entrance examination, and then we can also enter the Tsinghua University, and after seeing the title, it can give an answer. In fact, the exam is a very important method of evaluating artificial intelligence.
We are looking forward to maybe even in the future we may see a very good world of IoT, our mobile phones, our smart appliances, our smart home, our toys, our cars, all our robots, all of which are able to follow us freely. Communication, we can enter a revolutionary era of true machine recognition, thank you all!
Three-axis stabilizer is composed of pan axis, rolling axis and tilt-axis. With a gyro-stabilized gimbal system, it keeps stabilized or steerable horizon with automatic calibration to give you an unprecedented smooth shooting experience.
3 axis gimbal stabilizer can be divided into 4 parts, including Three-axis Smartphone Stabilizer, three axis micro dslr stabilizer, 3 axis camera stabilizer and 3axis motion camera stabilizer.
For different consumer, there are 2 kind of 3 axis gimbal stabilizer for them, which are consumer stabilizer and Professional Stabilizer.
Wewow focusing on handheld stabilizer is a technology company which does R & D independently. With Wenpod series product released, the company achieved the industry's praise and quickly became the leader of the smart stabilizer industry.
Our service
1. Reply to you within 24 hours.
2. Already sample: within 1-2days.
3. Shipping date: within 24 hours once get the payment.
4. 12 months warranty.
5. After-sales service, solve within 3 working dates.
If you have any questions, please contact with us directly.
Wewow appreciates domestic and international business relationship!
Three-Axis Stabilizer,Popular Three-Axis Mobile Phone Stabilizer,Professional Three-Axis Stabilizer,Handheld Three-Axis Gimbal Stabilizer
GUANGZHOU WEWOW ELECTRONIC CO., LTD. , https://www.stabilizers.pl