Here is a transcript of Professor Jerry Kaplan's speech.
Greetings from Silicon Valley, I live in California, I am very honored to be able to borrow Intel's invitation, this time to attend the meeting in Suzhou, thank you for your support, I am deeply honored to have such an opportunity to share with you the artificial intelligence to lead the industry change. Whether as an individual or as my business, we have been working with Intel for a long time, if there is no Intel, the PC industry changes will not even happen, today I introduce AI, referred to as AI. First, let's make sure what the AI is. Then let's introduce the history of AI development. What kind of stage is it in now? After these two backgrounds, we give you a brief introduction of the new business opportunities in AI field, and how it affects our various industries. I'd like to take this opportunity to give you some advice on how to better integrate your business with AI. Finally, I also want to focus on the Chinese market, how the Chinese government can help companies to better meet the AI triggered by the technological revolution.
First of all, I want to ask you one question: What is artificial intelligence? The vast majority of people talk about artificial intelligence, we all worry that the machine becomes too intelligent, even evolved to replace the human ability, will steal our work, resulting in a large number of unemployment, it will even directly control the whole world, why think so? Because what we see in the computer, such as Terminator. The rest is so scary, I won't let it go. In any case, the movie is very scary, but I think AI is very bright. The movie is not really real, we are in these science fiction and movies in the view is beyond imagination, and is a bit excessive, a bit of imagination. Think about today's robots, the most skilled engineers in human beings, and compare the robots we've developed, to what astonishing achievements can be achieved in some of the tasks of the U.S. Department of Defense. Can you see what this robot can do? Today's robots are really stupid. But if they can be optimized and upgraded, can they really develop into intelligent robots like the Terminator? Can they really rise up against humanity and eventually rule the world? The answer is No. Because they do not have their own desires, they do not have their own ideas, they will only do the human people to set up a predetermined task. They are presented in accordance with the prescribed procedures, even if they are to achieve the goals of the human race. Back to the first question, what is the definition of artificial intelligence?
An official definition of this is to research and develop computer systems that mimic human intelligence, such as visual perception, speech recognition, decision making, and language translation to perform tasks.
However, this definition I think is very incomplete, we often use the computer to do a lot of tasks, more than human efficiency is much higher, we build the system, but also hope to be far more efficient than the people itself, not simply to copy people's ability. Ai, in short, does not allow systems and devices to think like humans, the nature of artificial intelligence in my view, is automated, not intelligent, this is my interpretation of artificial intelligence and understanding. The new generation of equipment can always improve people's efficiency, whether it is intelligence or physical exercise, and related work can be significantly improved. They perform these tasks better, faster and cheaper than humans. Otherwise, we will develop these devices will be useless, but to the future, artificial intelligence can help us to do more. It will continue to automate, not just because machines are smarter, but machines are a very important tool developed by humans, and they make humans more valuable, freeing us from jumbled, stylized work.
Let's take a look at the history of artificial intelligence, how it has changed from its inception to the present, and why has the AI changed?
The development of artificial intelligence, from logical inference to machine learning
Artificial intelligence was actually born at the Dartmouth Conference in 1956, and some scientists held a special discussion during the meeting, and at Dartmouth we witnessed the birth of this artificial word. The participants at the meeting thought that the key to intelligence was logic reasoning, so what was the basis of logical reasoning at the conference where Ai lay the groundwork? We assume that Confucius is human, 2nd, everyone will die, all mortal, this logic is placed in the computer program, The use of such a statement to make a conclusion that Confucius would die is a very simple example. This is a logical approach, it has a lot of applications, and there is still a lot of impact, including giving us the right direction of driving; Using these logical methods in warehousing for better warehousing, inventory management, and at the same time, for computer chips to meet its specifications, also requires this part of the logical reasoning that Intel is also a leader in such technology. And in other areas, it's hard to solve these logical reasoning abilities that were presented at the 1956 Dartmouth conference, including translation between words, semantic understanding, computer vision, and so on. What are the similarities behind these questions? What's behind it is that a lot of it is very messy, unstructured real-world data is no way to be well explained, including the sounds we mentioned, including a lot of the image semantics, so a different approach is needed to enable AI to develop better. And now we use this word, machine learning to represent.
Behind the machine learning is a series of applications, including software technology, including the selection of different use models, cases, including a large number of cases to extract the model. Data collection in the same field, if the amount is large enough to find some models, but you can make full use of those models to predict or classify data, especially for new data in the same field. So on the other hand, machine learning is also a very good example of a better prediction of your future, assuming that the machine learning program is used to identify images, identify objects, and images include pictures of different cats and dogs, and it can find correlations, This correlation may be anything like the correlation of stock prices, which he found to be used to make stocks, including sales of securities, and so on. We are experts in this area, and we all know that logical reasoning and machine learning are two different areas of artificial intelligence, two different paths.
But there is another problem, why is it that the logic of logical reasoning has become so popular in the past few decades that it is now learning to be a big deal? If there is a simple answer behind it, there may be different stores, networks, calculations, and data that are undergoing profound changes. The volume of their changes is not a little, nor is it a lot, and I want to say that this word is a huge amount of change in the day, and this change is also the possibility that many may become more infinite. Looking back over the past few decades, in speed and memory, almost every year, its ability to double, which means that the current computer, now the ability than in the past 30 years, should be more than 20 times, if see his index, is 2 20, the volume should be more than 1 million, So even beyond our intuitive analysis and knowledge to understand this huge mass, 1 million represents the speed aspect, the speed of a snail's walk and the Space shuttle, what is the difference if you put it under the same order number? 500,000. Looking at Apple Watches again, Apple watches its computing power and the traditional American space program, which was the earliest, much more computationally powerful than the 1965 lunar landings, and the same changes also include digital data, and the volume of mass is enormous.
Why does this change the direction and method of the whole AI? First method, artificial intelligence of the reasoning method, he only a small amount of data can complete a lot of work, as long as the input is very little facts, can have very good reasoning and deduction, to draw a lot of useful user cases, and this for our existing technology, In particular, for 20, in 30, when the technology was available, this approach was undoubtedly effective. By contrast, before 30, computers, or machine-learning algorithms, lacked data, and even with this part of the data, the storage-processing power behind it would limit the direction and speed at which it could develop. So back in 20, before 30, this is why the logical reckoning is dominant.
At that time, people were not very concerned about machine learning, and over time, the situation had reversed. Machine learning what it can bring is a very messy, unstructured real world problem because the data is much larger. He needs a lot of computing power, a strong cloud storage capability, and it takes a lot more data, the more data, the better, the more accurate, and therefore, machine learning is a very good match, especially in a world of very high data densities, and in the era when we are about to enter 5G. This is the history of artificial intelligence.
Ai some important industry applications
We also saw a series of new and different applications, is closely related to AI, but it is important that we also need to understand these core technologies that drive the development of AI, and they have a lot in common, and all the AI-related applications, whether it's self-driving cars or product recommendation engines, or whatever, Network security algorithms and technologies, such as those that prevent us from being attacked by a network, actually represent the collection of data. Sometimes, this data comes from a traditional database that has a lot of transactions and may have a lot of records. Sometimes the torrent of data or data comes from a device terminal such as a real-time camera, microphone, or IoT. Here, what does machine learning mean for people in the industry and business people? It depends on your niche industry, artificial intelligence through machine learning has a wide range of applications, I give you to choose a few.
The first is very obvious, that is, representing a new generation of robots, it can be in the space of some people, work together with human, safe work, they actually in the factory has begun to be used, do some more repetitive, simple operation. This is because we now most of the robot's ability to perceive the environment is limited, there is no way to achieve a strong adaptability, so can only be used in some tightly controlled environment. But with the introduction of artificial intelligence, we are now able to give the robot a human eye and ear, it can have a strong sense of perception, and the surrounding environment to make interactions and reactions, even as some people and animals respond. These new generation of flexible robots have the ability to collaborate with people in real-world environments because they have strong adaptive capabilities and strong adaptability and responsiveness to their surroundings. These industries, including manufacturing, construction and transportation, will be subverted by a new generation of flexible robots. For self-driving cars, Sandra Rivera, a new representative of this generation of new flexible robots, is one example. In the next video we can see the driver's perspective on the left, on the right is the car sensor and the car camera head to see, you on the road, if someone appears he will automatically stop. For self-driving cars, it is very obvious in the field of the application industry, but what is very important is what? I think it's very important to have an automatic truck or convoy here. On the freeway, it may be much simpler and more regular than the very busy city road. So the use of self-driving trucks on highways may come sooner than the self-driving cars driving around the city roads, in our neighborhood. This is an example of an American, self-driving car, followed by a beer, and in 1 million miles of transportation, he had no human pilot intervention, the human observer, who was able to drink beer while he was driving. Behind the scenes, we can see that it frees the driver itself, and it costs less, and it can deliver almost every shipment.
Then we continue to share how flexible robots change factories. Consider the current plant or workshop, and the result is catastrophic if your device doesn't appear in the right place at the right time. The reason is simple, because most of the machines now do not have the ability to perceive the surrounding environment, to make rapid adaptive response to changes, but the flexible robot can perceive the surrounding environment, in a new way, different ways to make adaptive adjustments. More and more factories, mutual content, will be very complex, and it can have and human thinking as intertwined role. In this case, the automobile assembly plant, for example, based on the location of the parts, and a variety of different variables to use these tools, converters, for better perception and flexibility of production. This change in all the factories is going to happen, and we want to imitate it when we design it, which is how they operate if they are people and not machines. The assembly line of the past, these organizations are all we want from 100 years ago, when Henry Ford, the inventor of the automotive industry, founder, the founder of the Ford industry. At that time, the assembly line of cars was like this picture, the assembly drawing is made up of one station to the other, each of which is equipped with an assembly specialist to perform their tasks, the workers who must know at what station to install, and the other managers, to check the assembly line, to see here, to see where , to see what the problem is, in each of these processes is based on the different position of independent open, everyone is able to see, are external, can see what happens, now most of the manual operation is replaced by the machine, which is the modern assembly line, it is imitating the human operation before, This is like what Henry Ford did a century ago. Now it's not just that there is a person who has a very professional ability, also needs to have the equipment, has the flexible robot to do some special task in the specific location and the scene, in fact if through AI's introduction, the AI's program, Able to use these robots he can move, not just fixed at that station to do this thing, he can move between the equipment, if the problem, other devices can continue his operation independently. Unaffected by part of it. If you put it in a more complex, but very efficient organization, we can consider Amazon, Amazon is also the largest electricity dealer in the United States, retailers, they also began to use the robot Automation warehouse. In a classic traditional warehouse, the goods are put together, to find it, must be very clear, how to put it, but now the latest Amazon warehouse, these different things may be more than each other, this part of the shipping area is often shipped. may seem very messy, because they have different shapes, different stacking ways, different numbers, it's really hard for people to tube, but these devices using AI can achieve fast access to goods through a group of flexible robots, which greatly improves the efficiency of our work In the future and now there are only robots left in the warehouse, he will be able to bring these factories to a high degree of automation, he would be the future of manufacturing. Be able to use the technology of artificial intelligence fully. As a result, these plants will still be significantly different from the modern, automated plants that can be used in different devices, at different stages, including the use of these sensors, including the legions of intelligent robots that carry sensors, and they have the ability to identify themselves, Be able to transport, deliver, and move goods in a timely manner where needed. Therefore, the use of flexible robots will be very extensive, it will be in the construction industry combined with artificial intelligence technology can also play a very large role. This is a brick man, a robot, especially a robot that senses the surrounding environment, can make bricks very precisely, build a very solid wall, and its location is so precise.
Some other artificial intelligence applications are related to customer service. We know that this will affect many of us, and I choose one of them, the financial services industry as an example. We know that the financial services industry has a big challenge, how to respond quickly and efficiently to customers ' problems and needs, but in the next few years, we are very likely to have a special conversation, the machine will be able to answer the customer's request for financial management, to bring very advanced natural language processing, including semantic recognition, Language recognition. Perhaps through you say, for voice recognition, you speak out of natural voice processing, is also the introduction of a highly automated chat robot behind, can handle the response is very complex customer issues, including some intelligent chat robot in the micro-letter has been used. But it develops the speed of automation, the speed of maturation can be increased through the introduction of artificial intelligence technology.
In addition, perhaps driven by AI, the technology that makes us feel very excited and exciting is AR augmented reality field. It brings a translucent image to the AR glasses, wears this kind of glasses, it can scan your surroundings, he can introduce a very good rendering of high-definition map, augmented reality technology, will bring the future of the game. The next generation of games will be highly integrated with objects around you, and here is a recent Microsoft show, here is what they see after wearing this glasses, it is the real environment behind, that is, you are in the environment, here, he put on this glasses, He can have a good interaction with the wall around you, even though the home has real furniture, but the game is like immersive, in that game environment, may personally feel a bit scary.
Not only is this industry, AI most far-reaching change is the big medical industry, what can help is the patient, doctor, and it can even take the place of a doctor, and we might think, because it's a very specialized job here and it seems to be hard to automate their work, that's not the case. For example, for imaging doctors, doctors in the imaging field spend most of their time watching movies, like X-rays, MRI, CT scans, they spend years learning to explain, to learn how to watch the film, to see whether the black spot is benign or cancer of the tumor, only the imaging experts can make such a decision. This is important and relevant, however, because we know that cancer is a very important cause behind China's death rate, and so is the case in the United States, where imaging and radiology physicians, who have mentioned this day, may have died of this disease in a year or more than 600,000. Including CT scans of doctors, whether they can diagnose well is not the presence of cancer is critical, I want to say that in China and in the United States, we are very, very lack of qualified imaging doctors, so their workload is too large, they may, now some people may be due to technical problems and even misdiagnosis, Included in the United States. But if they use machine learning to quickly diagnose many cancers and other illnesses, it is the work of a video surgeon. The technology has been largely adopted in many Chinese hospitals and clinics. The changes in the health industry, not only in the field that we have just introduced, but also the way we measure health management can change dramatically, and we all know what a IoT is, but what you might not think of is that we are one of those things. More and more sensors will be worn on us, these sensors will detect our health problems and vulnerabilities in real time, and if there is a problem, we will immediately signal to you, or directly inform your doctor, in the future is not just equipment, and even our body can be linked to the Internet, Today's health industry is to be proactive when consumers need them. When you feel sick, you go to the doctor, this is the current set of processes, but in the future, the situation will change, we will be the emergence of a new service provider, we can track your health in real time, through this predictive analysis ability, we can directly identify some potential health hazards. These service providers can make personalized adjustments to your health status, such as your heartbeat quickly, you have to take a break, or give you a more personalized treatment regimen. For example, can you eat a watermelon for dessert or no more? He can better detect patients, urge you to take medicine quickly, or when your symptoms become worse, such as a possible heart attack, which we can do in the medical field. The relationship between patients and medical experts will change dramatically, which will greatly subvert the service model of the healthcare industry, and we will have a more exciting medical profession.
In the last part of my speech, I would like to introduce to you one of the most promising areas of my life, which is the application of AI in retailing, and how much it will overturn the retail apparel sales industry. In fact, in the apparel retailing industry, the biggest problem is that many people's taste is poor, buy clothes are ugly, no matter you do not realize this, so he also knows his own this problem, so said in the acceptance of a friend's opinion, I do not dare to buy this looks good clothes. Of course, your friends or sales, the advice they make is not really good. These consumers want a stylish expert to give me a useful view, and a new AI application is already on the Amazon. How are you going to use it? You can try different clothes, take a self-portrait of yourself, and then put these self-portraits upload to this application, this application will tell you, which clothes make you look more beautiful, you may think it is Arabian Nights, estimated also useless, but according to our actual test results, we found that The use of calculations is more than the opinion of the fashion experts.
This can be said to be a huge change in the retail industry, anyone can get the advice of fashion experts, they can also be more confident to buy their favorite clothing products. This is a very simple way to combine artificial intelligence with retailing, but I think it's very interesting and it's going to subvert our way of life. AI does not just change the existing industry, it is also constantly to generate new business areas. I would like to finally spend a little bit of time to look at the future of innovative equipment, how can it radically change our way of life?
Today's existing equipment needs our full attention on the device itself, I don't think it's a good way to interact with future devices, we'll solve this problem. Devices will become more accessible, more invisible, more hidden, and we'll wear this very small headset, or this transparent contact lens. , even this device can be directly integrated with contact lenses, it will not interfere with our daily life, they will be at any time behind the provision of relevant services, this future computing equipment will greatly improve our communication and communication with the environment. These new sound interfaces are already in place for existing devices, it can provide us with very useful information, including your turning on the road, or receiving an email, and so on, and they can do more, they can be electronic personal assistants, and he'll tell you what to do with your ear, For example, don't forget to take the key, he at the table under the napkin, or you have 3 days without your wife, should do something to make her happy. For example, your blood pressure is a bit high, you should go for a walk in the afternoon. Look at your contact lenses and your contact lenses will be better. Suppose we are going to have a meeting, your eyes will automatically identify each person, everyone, do not need to put such a long name on the table to do the introduction. It can also be a very good life assistant. It will automatically tell you that the person sitting opposite you will have a hobby, such as it likes to ride a bike, and then this man's daughter is your classmate, then he sent you a birthday wish last year, you can express some gratitude to him today.
In addition, these personal assistants can also help us to be more athletic, tell you where the tennis is flying, where it is, tell you to go out for dinner, it will even affect our intimate life.
How does AI better integrate industry applications?
Below I also want to provide you with a little advice, how to make AI and their business more perfect integration? Without a doubt, machine learning is a way to extract effective information and to find data that is truly valuable to itself in the ocean of data, But applying it to your company's application scenario is another important step, and in order to do that, we have to search through a lot of data, and the more data you have, the more successful your machine learning application will be. The value it can bring to you is more pronounced. We may have a lot of different types of cameras, your warehouse, bank counters, most of the cameras are used for security, and most of the data collected is already wasted. We can use this data to analyze the behavior of customers, we can directly tell employees where to go to serve customers, but also a lot of energy savings, save resources waste, in short, machine learning can make the camera such devices collected data to play its greatest value, can also be better for customers to achieve personalized services, data is indeed the future of oil.
In addition to my second suggestion, to be cautious, carefully chosen, we should not be a little bit, but the first must do a very good pilot, artificial intelligence is not a panacea, we need to take a moment to see how this technology is more in-depth integration, artificial intelligence must improve their ROI, What we need to do is not just attract eyeballs. I suggest that you start by testing some potential applications, build prototypes, see what value they can bring, and then choose whether or not to scale.
Finally give everybody's suggestion, is not too anxious. If you are too anxious, you may overlook the problems that may exist behind this, and now there are fewer experts and talents in the field of artificial intelligence, but in the future it will be greatly improved and we will develop more software tools to make the AI system easier to deploy. Even in our conference room today, there are similar devices and technologies, and in the future, new machines will require better training to achieve the same AI efficiency. A lot of equipment is very expensive now, but the cost will quickly fall, our talent problem will be solved. Tools will also become more mature and more universally available.
Some suggestions on the development of artificial intelligence in China
Finally, let me introduce the Chinese market to the AI how to prepare, and how can the Chinese government to meet the AI revolution?
China has a very good potential to become a leading country in the AI field. One of China's great strengths is the ability to achieve very good synergy, and everyone can work towards the same goal. However, China is too big, the data is very much, how to do the data and the training of the machine system become key. With the support of the Government, we can better encourage the sharing of data, to the aggregation of such a large database, this is a way China can be considered, the use of AI data to grasp a very advanced technology advantage. The government can also provide other help, we can optimize existing infrastructure and laws and regulations, perhaps AI companies to accelerate the pilot of their own AI devices and applications, many infrastructure is not in line with the future changes, while, like intelligent transportation, we have to prepare for the future of automatic driving and automatic taxi. In the future we will be on the road to the design of traffic to conform to the modern way of life, sidewalks and robots must be separated from the road, so that the two sides will not have conflict.
Finally, a brief summary. I believe that artificial intelligence will make our future brighter. In short, more flexible equipment will take on human physical work and more stylized work, it will make people's intelligence in the future more valuable and valuable, in addition, artificial intelligence will help us better meet the challenges. Benefiting from this digital world. It can also automatically find the most valuable information to help us make more correct decisions. Nevertheless, in the final analysis, the good future is to be created by people, the machine is only an important tool to achieve, how to use these technologies rely on us, I also very much look forward to seeing the arrival of that day, I hope that we can collectively brainstorm, find a more innovative way to use artificial intelligence to serve our work, Create a better world.