With the development of artificial intelligence as a national strategic layout, both entrepreneurs and investors are looking forward to taking the lead in the wave of artificial intelligence, designing and manufacturing smart chips, and identifying, processing and communicating information such as voice, semantics, images, and actions. Feedback, which areas will become the next flash point for artificial intelligence? What is the difference between the method of processing information in the cloud and the terminal's artificial intelligence? What kind of opportunities and risks will the development of artificial intelligence bring to society and humanity? On the afternoon of December 12th, Sohu Ventures SoPlus invited five entrepreneurs, investors and experts in the field of artificial intelligence in the field of artificial intelligence, Chen Zhongmin, vice president of technology, Shenzhen Jianchao, CTO of Triangle Animals, and Cheng Jian, a researcher at the Institute of Automation, Chinese Academy of Sciences. Hong Taizhi’s CTO Qian Chen and director Qualcomm of Qualcomm Venture Capital conducted an in-depth discussion on artificial intelligence in Beijing Sohu Media Building.
Hong Taizhi includes three parts: workshops, funds, and industrial parks. Hong Taizhi’s manufacturing workshop focuses on hardware incubation and product incubation. Its CTO Qian Chen uses an example to illustrate the essential difference between artificial intelligence and people: a teacher teaches a group of students and students present There are completely different results of lectures, and there are sciences in the liberal arts. Machines are not the same, and the machines are all taught the same personality traits. The diversity of people is enough to distinguish the essential difference between artificial intelligence and people. Moreover, while artificial intelligence is coming, human creativity will continue to develop. However, the creativity of artificial intelligence is based on the logical analysis of its experience.
Regarding the impact of artificial intelligence on society, Qian Chen said that the emergence of a new technology will lead to new social phenomena. This social phenomenon will lead to legal responses. The three parties will adjust each other and the efficiency of adjustment will be crucial. If efficiency is good, it is a If the balance is not good, either the technology is not suitable for social development or the law needs to be readjusted.
In many areas of artificial intelligence, Hong Tai Zhi Zhi is currently more optimistic about artificial intelligence in the field of industrial intelligent manufacturing, natural customer interaction scene is too complex, the data is not easy to do. However, in terms of intelligent manufacturing material detection, the scene is very fixed and data processing is relatively easy. It also saves a large number of IQC incoming inspection engineers and shipping inspection engineers in the factory. The image inspection is a big pattern in smart manufacturing.
Shen Jian Technology is a startup company that is longer than neural network accelerators and chips. Its vice president of technology Chen Zhongmin shared the differences between strong and weak artificial intelligence: strong artificial intelligence, capable of thinking, reasoning and even showing some creativity, more like people. In fact, this matter is scientifically speaking, so far no technical level has been able to achieve this. On the contrary, artificial intelligence that we can actually feel is actually a category of weak artificial intelligence. It just means letting the machine replace people to do something by analyzing and fitting people's way of thinking.
But does the AI chip that Shen Jian Technology is good at really just need? Chen Zhongmin introduced that everyone pays more attention to the chip's acceleration function for AI. The main reason is that traditionally we used CPUs to some extent it is very general, and it sacrifices some requirements for computing performance for commonality. But after the emergence of AI, its algorithm is relatively single, and many people find that if I abandon the processor today to do this thing, I design a hardware structure that is very much in line with the needs of the algorithm, and at the same time we can make the CPU not spend a lot of chip area This matter only took a little bit. Most of the CPU stuff was useless to us. We reduced that part. The GPU is similar in principle. Parallel computing is very good. So the support of the algorithm for the algorithm, especially the support of some ideas of the neural network has become very good. From the chip's point of view, not because of this chip technology has made some breakthrough progress, but that if you only serve a single algorithm, this chip can be done very special, stacking a lot of mana, relative to the traditional GPU processor performance Better performance. Algorithm changes, scene changes, leading to the discovery today must make some improvements in the hardware in order to do better. We have abandoned the commonality that CPUs are pursuing, and only do something for parallel computing. This will yield very good results. From the point of view of computing power, in particular, AI chip companies do some real-time experience for users, and special hardware devices will become very important. This is the basic logic for the development of AI chips.
Focusing on technological innovation, especially the high-tech venture capital executive director of the cutting-edge technology, Mao Wei, shared the evaluation criteria of technical value from the perspective of investment: Before the value of technology itself is actually dropped, the technical value itself is difficult to assess and test. Technology is good technology. Then what kind of industry can the technology help or what kind of person can solve the problem? Can not solve practical problems, can not generate income, but also can not create profits, such technology is commercially limited in value. "For entrepreneurs in this direction, don't go behind the scenes and try to find out more. Try to figure out where the technology is going and how to use it." The value of technology itself is limited, only if the technology finds a reasonable application scenario. And it really generates value in the supply and demand.
“Across history, you will find that all of our good inventions actually do the same thing. It is to make us more lazy. We don’t want to walk, so we invent the car. We want to go faster and further. So we invented the plane, and now we don't even think about it, we develop artificial intelligence, and then everyone slowly becomes lazier. Today's artificial intelligence is weak artificial intelligence, and I very much agree with this view, that is, our current Artificial intelligence is still an intelligent solution for a specific field or even for a specific application, after repeated training with a large amount of data, such a weak artificial intelligence creativity is limited, but we humans have a huge difference with it. The basic idea of artificial intelligence is a regular solution that is based on repeated training after a lot of calibration data for a specific problem or for a specific application. However, our human ability is based on non-repetitive training of small data to derive its regularity. This is The essential difference between the two."
Language and text are the advantages of triangular beast technology. When talking about the two-year experience of entrepreneurial ventures, Delta CTO Yuchao deeply felt that although the triangular beast technology is good, the business model of finding the scene and finding the landing is particularly difficult. The distance to the real landing is still quite large." At the same time, artificial intelligence is not a new alternative to human evolution, but a supplement to the role. Various robots appear in science fiction cartoons and science fiction worlds, and their roles are more complementary. For example, when Vikings appeared at Conf's home, Conf's father and mother did not disappear. The classmates did not disappear. They only added a 24-hour robot that could accompany him to help him spend his childhood. He did not replace someone to do this. Therefore, even if artificial intelligence develops to the stage of strong artificial intelligence, it still has the mission they were given when it was developed, and it becomes part of human society, and there will be no case where people are completely replaced by artificial intelligence.
As for the impact of artificial intelligence on society, Min Chao is optimistic: “Every iteration of the technology will promote more possibilities, make people's lives better, or have a higher degree of comfort.” Early through hunting and catching The fish go to live, but with agriculture, you can eat steadily and start to create more arts or create better jobs for human development. This is actually a promotion for people. Artificial intelligence is still helping people solve practical problems and help the development of society.
As representatives of the scientific research and academic circles, Cheng Jian, a researcher at the Institute of Automation at the Chinese Academy of Sciences, expressed his understanding of the concept of artificial intelligence: artificial intelligence is not specifically defined, and there are many different opinions in the academic community. Some people say that artificial intelligence thinks like people. Act, this is artificial intelligence. Some definitions of artificial intelligence, such as the ability to simulate people and extend the intelligence of the extended person, are accurate and very extensive.
Cheng Jian believes that artificial intelligence will not surpass humans. In certain areas and areas, artificial intelligence has indeed surpassed humanity and may surpass people in more areas in the future. But transcending people is not terrible. People develop more time and energy to do more at higher levels or in other areas. Since there are two artificial intelligences, one is artificial and the other is intelligence. It must be based on human simulation or in the range that people can think and reach to simulate such intelligence. Machines that transcend people will always surpass only on specific ones, but they will never replace people. Moreover, the time required will be very long and may gradually replace or exceed people in many fields. However, people can develop more new things, more interesting things, and challenging things to do.
Cheng Jian also expressed his optimism about the future of artificial intelligence: “Actually, AI chips are quite promising. This is the first direction that I am optimistic about. I came out of the Institute and I can feel it myself. Now that artificial intelligence is Fire, many people do not understand artificial intelligence techniques and algorithms.From this perspective, the future of artificial intelligence education is particularly hot. There are two kinds of education, one to teach everyone to learn artificial intelligence technology, methods, and in the education process. Artificial intelligence technology is used in these two kinds of future is very promising."