Advice for entrepreneurs: grasp the conditions of artificial intelligence

First of all, we have to say that entrepreneurship is a loser. Why is artificial intelligence chip a "noble game" that entrepreneurs can't afford?

In the past many years, there has never been a year like this year focusing on artificial intelligence.

Li Yanhong repeatedly mentioned artificial intelligence - whether it is willing or not, artificial intelligence has come;

Then, during the G20 summit, Jia Yueting took the initiative to expose their actions on artificial intelligence on Weibo – “I look forward to seeing the world’s first LeEco artificial intelligence (LeAI) eco-phone available as soon as possible, and I will see you soon.”

At the same time, Liu Qingfeng, chairman of the board of directors of HKUST, who attended the B20, said: "If I don't consider the trend and impact of the artificial intelligence industry when discussing global economic governance, I think it is totally unimaginable."

And the discussion boom in the technical field is directly on the peak. It is a paragraph of Xi’s speech in the G20: “A new round of technology and industrial revolution with the Internet as the core is ready to go, artificial intelligence, virtual reality, etc. New technologies are changing with each passing day, and the combination of virtual economy and real economy will bring revolutionary changes to people's production methods and lifestyles."

Looking abroad, Google has also begun to adjust its strategy, from mobile priority to artificial intelligence.

Although at the beginning of the year, when Alphago fought Li Shishi, the media had already interpreted various kinds of artificial intelligence, but many people still have such questions: Will artificial intelligence replace humans? Where is the real threat? Does it have emotions? Is the future smart, or cloud smart? In addition, what are the conditions for artificial intelligence to land? At the commercial level, what is the business model of artificial intelligence? How should entrepreneurs in this field start a business?

Note: Last month, Alibaba iDST researcher, IEEE Fellow Hua Xiansheng (flower name: Fang Guang); Alibaba iDST director Chu Min; Alibaba senior data mining expert Wan Wanli (flower name: mountain view); Alibaba Cloud Research Center Director Tian Feng (flower name: wine road) came to a "boiled wine" on artificial intelligence.

Will artificial intelligence replace humans? This is just the language of the scientific community.

It is often said in the scientific and technological community that artificial intelligence is to replace human beings. But such remarks are seen by the three artificial intelligence big coffees. It’s just that the words are not surprisingly endless – they attract attention with sensational remarks, because all the robots seen today are still machines, and they are still The difference is very far.

The three teachers explained from the following perspectives:

Technology surpasses humanity and it has long existed;

The limitations of the machine: can only do the ability of logical thinking, no perceptual thinking, it is too early to talk about transcendence;

The relationship between human and artificial intelligence: it is an extension of the five senses of human hands, eyes, ears, etc. Its ultimate role is to serve us.

The first point: technology surpasses humans, such as the abacus that appeared thousands of years ago (faster than mental arithmetic), the calculators that were decades ago, and the deep blue that defeated humans on chess 20 years ago. Intelligent technology has also been used in many places for decades, such as moon landing technology and industrial production lines. The reason why today will become a popular topic, mainly today's computing power, algorithms, machine learning and data capabilities, so that problems that were previously difficult to solve have been solved today, so they will be interpreted too much by the public.

The second point: string together the examples above, and you will find that they are good at rules and rules – emphasizing the ability to think logically. Therefore, in some things that can be written as rules, artificial intelligence will exceed humans. However, in sensible thinking such as literature and art, it is not an objective law. It cannot be written in logic as a simple rule equal to two or five divided by two. In this case, talk about artificial intelligence to replace human beings. It is too early.

The third point: the vast majority of people doing artificial intelligence, his goal is not a substitute. It just looks like a person somewhere, but it doesn't really have human emotions and wisdom. All the technologies, including industrial technology, are generally an extension of human senses and limbs. Behind these things are people who dominate. "The development of these technologies is ultimately to serve us. It is to liberate our hands and time within a certain scope, instead of replacing the whole process of thinking." Chu Min said that the future must be people-oriented, supplemented by machines. What is the specific, not very predictable, "I believe that coming back after 10 years is not something we can imagine today, because there will be many changes here."

On the whole, human and artificial intelligence are the relationship between strengths and weaknesses, artificial intelligence is strong in the integration and summarization of complex data, and people are strong in intuitive feeling. Chu Min said that it is an extension of our hands, feet and eyes, not an opposition. The future is actually just a difference in division of labor. Through the machine, our life is very convenient and simple. In addition, because artificial intelligence is driven by big data and limited by the data field, artificial intelligence will present two opposite phenomena in the future: one is more and more intelligent, and the other is still a bit silly in some areas.

Someone here may be concerned about which jobs will be replaced by machines. In the era of industrial revolution, physically intensive was replaced by machines. Chu Min pointed out: "In the era of artificial intelligence, the first thing to be replaced should be human-intensive, and the second is partial brain-intensive." Fang Guang said that as long as it is in the data Intensive, computationally intensive, there are examples of areas that can be found, and machines can surpass humans.

Is the threat of artificial intelligence coming from people or machines?

There are many celebrities who make artificial intelligence that will lead to the demise of humanity. What are their intentions? A common understanding among the three teachers is that the threat of artificial intelligence does not come from the machine that is given the so-called intelligence, but from the person who made the machine.

Mountain View indicates that the development of any science and technology is a double-edged sword, and it may all be counterproductive. For example, someone puts malicious code into it, even if it is a simple robot, it can cause great harm to people. So, like today's anti-virus software, there will be similar appeals for artificial intelligence in the future.

In a certain aspect, the technology that transcends the limits of humanity is not at fault. It is like searching. It is a great transcendence. Going to the library to turn a book, can one person look at a good book in a year? How much information can I find? "The essence of technology is still convenient. Because the robot has no sense of autonomy, the so-called threat is the logic behind the people doing bad things or someone teaching the robot to do bad things, not the robots themselves doing bad things." Don't worry about this, at least don't worry about this in the time you can see it.

If you talk about how to prevent it, the recommendations are as follows:

1. There are fewer bugs in the logic code;

2. Use the law to constrain the people behind artificial intelligence.

Spontaneous emotional intelligence is a particularly difficult thing

"To be emotionally intelligent, we must first let the machine understand the emotional world of people. But the human brain has tens of billions of neurons, and each cell has active and inactive states (0 and 1), one day. The amount of data is about one hundred PB. If you can calculate the cytokine activity map of the whole brain nerve every moment by calculation, restore the state of each of our emotions - in what state is happy or unhappy, Which cell elements are activated to cause such an emotion, and what causes it? If this problem can be solved, at least the channel of our emotions can be understood from a cognitive perspective, and what its mode is. ”

Only after understanding this working mode can it be possible to build an emotional robot. How to make a machine have a spontaneous emotion, Mountain View said: "So far, I think it is still very difficult. At present, I can only do the first step first, first understand the generation of emotions."

He also said that in terms of emotional understanding, today's big data can already be partially achieved. Although it does not show the activity map of the cranial nerve, it can simulate what the stimulus data is, and then guess the deep level through the representation. , you can achieve some emotional functions, such as listening to music. “It’s just to do something quantifiable emotionally, then infinitely close, but infinite approximation does not mean spontaneous creativity.”

Chu Min believes that emotional intelligence is divided into two levels, one is to let the machine itself have emotions, and the other is to let the machine understand people's emotions, these two are different. It is feasible to let the machine understand the human emotions, but it is still too difficult to make the machine emotional. For example, the so-called companion robot, the companion robot does not teach the robot to have emotions, but to let it grasp the emotions that it needs to accompany the object, and what the object needs to do in different emotional states. These things are completely Can learn.

Is the future smart, or cloud smart?

There are two trends now. One is that cloud computing companies do smart things on the cloud, and the other is that startups do smart things, such as cameras, mobile phones and other smart devices. The future trend is the end. Smart, or cloud smart?

Fang Guang analyzed that the advantage of end intelligence is agility, no need for voice interaction, and immediate calculation results, so for some applications with high real-time requirements, it is necessary to use the end, such as automatic driving, security, and so on. Its shortcomings are limited by its ability to calculate, but also can not feel the other end (if it can, it is the cloud). The advantage of the cloud is that it has powerful computing power, and it can combine these points—the collected data at each end to analyze and judge. This kind of intelligence sometimes transcends human intelligence.

Fang Guang believes that the relationship between the cloud and the end is interdependent: the technology of the cloud needs to be presented through the end, and the end is the entrance and exit of the cloud data. He pointed out that after N years, perhaps the concept of cloud and end will be gone. When the bandwidth between the cloud and the end is not a problem, the end and the cloud will be integrated.

"Intelligence is at the end, wisdom is in the cloud." Mountain View said that when local intelligence gathers together, it can become a great wisdom. For example, if you capture the camera lights of a deck car and a red light in the cloud, you can do a lot of things. This is where local intelligence becomes a global wisdom. The end and the cloud must be connected together. The ultimate realm is that you have me, I have you, mixed together.

Chu Min explained the relationship between cloud and end from another angle. Between the end and the cloud, only the division of labor is different. Some things are not suitable for the cloud because of security and privacy needs, so they can only be handled in the end, such as the smart sound in the home, can not pass all the sound to the cloud.

Artificial intelligence landing conditions

Fang Guang believes that the conditions for artificial intelligence to fall are: algorithm, data, computing platform, user, business model, these five conditions are essential.

Artificial intelligence is first built on a sufficiently powerful algorithm, but there is no data in the algorithm, which is also difficult for a woman. In addition to algorithms and data, a computing platform is also needed to support it, otherwise the algorithm cannot be implemented. In addition, users need to be able to drive technology development and contribute data. At the same time, a good business model is needed, otherwise everything will be difficult to last.

The feedback in the process is very important. Mountain View says that it is like cramming education. The teacher gives you the title. It is wrong, or is right. What kind of problem should be given to the next round? The process of adaptation is critical.

These conditions are true, and if you don't use Internet thinking, landing may not be as fast. Chu Min said that using the thinking of the Internet to string these five factors together, the iteration can be very fast. “Using feedback data to update the model at a faster rate, after such a positive cycle, the effect will get better and better.” Even if the algorithm is unchanged, as long as the data can be continuously fed back and optimized, one or two After the month, its ability will be much better.

How to commercialize artificial intelligence?

When it comes to the business model of artificial intelligence, Shimin said that it is not the business model of artificial intelligence, but the value of artificial intelligence application is greater in solving some people's needs and commercialization.

From the perspective of division of labor, Chu Min believes that the business model of artificial intelligence has two levels: “The first level is that a group of people provide basic technology behind, and then a group of people fall to the product – using artificial intelligence in specific application scenarios; One is a macro judgment, which is a very trending scenario in which there is a great possibility of making good results, which in turn drives some companies to store these technologies in their applications."

“One is the efficiency improvement of existing business processes, and the other is disruptive innovation.” Mountain View summarizes the business model into two. The value of the former is mainly reflected in the value-added. For example, after the production line uses artificial intelligence, the efficiency is improved and the overall income is improved. The latter is a kind of out-of-the-box, creating new opportunities, such as steam locomotives instead of carriages, today is the emergence of driverless cars, it is possible to subvert the labor-intensive taxi industry, mountain view that these are Possibly, "It may be a restructuring of a business model, selling algorithms, or selling calculations, selling money based on miles traveled."

In the business model, Fang Guang mentioned a very crucial point - whether it really solves people's needs. "Is this thing useful? Is it a solution to some people, or a lot of people." Fang Guang believes that only by identifying this application, it is necessary to solve the problems of technology, data, platform and users. It is possible to truly become a commercial application.

Advice for entrepreneurs in the field of artificial intelligence

For the current entrepreneurs in the field of artificial intelligence, Fang Guang feels that it is still possible to find out what is the problem to be solved, what is good at it, and then look at the five elements mentioned above: algorithm, data, user , computing platforms and business models, if these factors are more likely to be successful, if you do not have a factor, you must think about how to get.

“For an entrepreneur, a more precise positioning is very important.” Chu Min believes that the most important thing is to focus, not to be too big, try to use the ready-made things around you to do what you want to do the most. There are different levels of artificial intelligence. To make a scenario, technology does not necessarily need to be developed by itself, or it can use existing technology. For those who work in the technical field, it is necessary to make the technology valuable and have a threshold, and finally sell the technology. Each company needs to accurately position itself according to its own characteristics.

Mountain View suggests that you don't make artificial intelligence chips. "This is a noble game. If you fail a few million dollars, you will burn it. This is not a capital for startups." He went on to say, One of the algorithms in artificial intelligence chips may be very bullish, but this technology lead may be diluted because it takes a long time to develop into mass production. If there is another thing that can replace it, then the whole value will be diluted at once, which is a high-risk choice for artificial intelligence entrepreneurship. Mountain View also said that for those who do the scene-business innovation, the role of weakening technology should start to stand up and grab market share, especially in our domestic competitive environment, otherwise it will be easily copied and killed.

At last

Moderator Tian Feng also extended a question: Is there a possibility of a super algorithm that allows artificial intelligence to discover patterns that humans cannot find, and how do people verify it?

Fang Guang said that this is possible. "Today's artificial intelligence is assuming a situation to test it. After artificial intelligence that enhances learning to a certain extent, it is really difficult to test whether it is good or bad." Chu Min said, you don't know the data is hidden. What, so you may not know what it has learned.

Mountain View believes that this super algorithm is very difficult to implement. He said that it is often difficult to prove the establishment of a thing, but it is very easy to prove its inadequacy and non-establishment. From this perspective, today's artificial intelligence has learned far from enough.

The three experts also made a message about the future of artificial intelligence, summing up two points:

1. The prospect of artificial intelligence is very bright, but the road is twists and turns - Long wait go. But as long as the direction is right, there is no need to be afraid of the road. As long as you keep going, there will always be a process of quantitative change to qualitative change.

2, artificial intelligence is the extension of our hands, eyes, ears and other facial features, its main role is to help us, so do not need to worry about the threat of artificial intelligence.

COB Light

COB LED par light for theatre, productions, TV studio, stage

Description:

COB200-2in1 is a professional theatre fixture that utilizes a 200W warm-white and cold-white COB LED with a color temperature of 3200k-6000K. Users are able to creat a customized color tempreature via a DMX Controller or set directly on the display menu. It offers a high-power light output with rich hues and smooth color mixing for stage and wall washing. The double bracket makes installing easily and versatile.


Our company have 13 years experience of LED Display and Stage Lights , our company mainly produce Indoor Rental LED Display, Outdoor Rental LED Display, Transparent LED Display,Indoor Fixed Indoor LED Display, Outdoor Fixed LED Display, Poster LED Display , Dance LED Display ... In additional, we also produce stage lights, such as beam lights Series, moving head lights Series, LED Par Light Series and son on...

COB Light Series,Led Par Light,54 Led Par Light,Par Led Lights

Guangzhou Chengwen Photoelectric Technology co.,ltd , https://www.cwleddisplay.com

Posted on