Editor's note: This article is to go out and ask Li Zhifei's speech in Sina C+, and Lei Fengnet (search for "Lei Feng Net" public concern) made an editor without changing the original meaning.
Ten years ago when we talked about AI, we would definitely not be on this occasion, but at academic conferences or on a project of the US military. For example, my own project is a project supported by the US military. Later, Google started to do translations and voices. Now that all people are talking about AI, I have found a particularly interesting phenomenon. AI has changed from a very tall, very distant, very academic field to a topic that the general public is discussing every day. The headlines I see on a daily basis are drones, AI, and autopilot. Many activities like this are organized. I personally receive some AI activities from the media, government agencies, and companies every day, so AI has It is an area of ​​great concern.
The AI ​​said so much. So far, as everyone else, AI is just AI in the media, AI on the podium, and AI in science fiction movies. The AI ​​products like sci-fi that we actually see in normal life are very few and have little influence on society and the public. So how did AI develop? So many people engaged in AI entrepreneurship, AI system can support a few years in the end? We should focus on these things next? Today I want to share some of my own feelings with you from your own unique experience. My sharing has two main perspectives:
The future of AIOne is the future of AI . Every day we see a variety of products, driverless, AI, Google Glass, these conceptual products in the video so that everyone has great fantasy and association with it, we every day hope that AI can come . There is one of the most famous arguments for AI, singularity, and there is a singularity university . I have said on many occasions that Kurzweil 's near singularity is a liar. He is a school between scientists, he is a futurologist, he describes the future, but has some of his own appeals, but this theory has received a lot of people's attention and respect, because it sounds particularly interesting, especially imaginative. I do not know how many people have read the book, "Singularity Approaching," and according to exponential growth, we can infer that the computer intelligence will be about the same as humans in 2025. In 2045, machine intelligence will exceed human intelligence.
There will be something that looks very scientific. It sums up an index law. What will happen now and what will happen in the future will sound very imaginative. It is not a good future for the machine to extinguish people. I don't know, but many people are willing to discuss the machine to kill people. He has the most famous sentence. AI is the last invention in human civilization because after inventing the AI, The entire human body is destroyed by the machine . This is a world that is very exciting or exciting. But in the end it was science fiction or reality. I didn't think it was worth criticizing before. But I think that so many people are interested in it and want to know about it. In fact, some of its viewpoints are more reliable.
Integration of human intelligence and machine intelligence
First, human intelligence equals the integration of bio-intelligence plus machine intelligence, which is his core point of view . This point of view is very reliable. There may be two parts of intelligence in the future. One is the intelligence given to you by biology, and the intelligence God gives you. The second is the intelligence of machines. These technologies are Americans to do. We can imagine that we don’t need to bring purses in the future. We don’t need to bring a bus card. Your body is implanted with a chip. This chip is your ID. You may not need it. With a variety of packages, as well as a variety of data for your life, this is very feasible.
Even we may give up some biological organs in the future, such as our legs. It may be that your leg is disabled or disabled before you change your leg. However, imagine that if you have a machine leg that does not harm your body, your ability is greatly improved, you can run very fast, and you can even fly. Can jump, someone can use the mechanical things to supplement human capabilities. This is technically mature and technically possible, but only if social regulations can be accepted. So, this may be easier than others.
Can Machine Intelligence Exceed Artificial Intelligence
Can machine intelligence surpass human intelligence? We can hardly refute that it cannot be exceeded because we cannot prove whether it can or cannot, and we do not know how to do it. As a scientist and engineer, you need to know how to do it step by step, or how to do it. But today we cannot see a possibility that machine intelligence can surpass human intelligence.
All of today's AI technology, whether it is speech recognition or computational recognition, will be very mechanical if you are familiar with its principles. For example, in speech recognition, we have some understanding of acoustics and human language, and how human languages ​​are created and communicated. We have a very deep understanding of this. Then we have sensors that speed up these data, such as speech recognition with a microphone, digitize it, and finally train a system. We can see that one is very familiar with the principles, the second is that we have sensors to collect data, and the third is the ability to process data .
Today's speech recognition, computation and recognition can be done on the level of partial perception. However, in the process of listening to me, there are many people who are thinking about it. They can think based on their own past experiences. Whether we can create a machine that can think or even exceed people’s IQ and let everyone collaborate is very difficult. For a very big reason, we don’t know how to digitize human activities. Digitizing life is very difficult. It is hard for people to imagine that I plug in a sensor and measure every word I say in my brain. This is very difficult. In the second part, even if you can digitize, you send a photo with a bandwidth. This is the process of digitization. As you can imagine, this is just a daily visual digitalization. If you have countless such ideas every day, you actually don’t It is digital. But because of these two points, we can't see today a model that can put people's intelligence, quantify it digitally, transfer it and produce a machine learning model. This is impossible at present.
The machine has a particularly interesting phenomenon today. It does not have complex reasoning, but at the same time it does not have simple intuition. People discuss AlphaGo's thought to some extent that for the first time, humans have begun modeling intuition, but we do not know how to model human intuition with machines or machine learning patterns. If there is no way to model intuition, it is often not possible with intelligence. Because we hear every statement, we do not measure it through ABCD, but intuition that arises immediately.
To learn how to do ABCD through machine learning, from the first step to the second, third, and fourth steps, all the principles of the computer are done through this model, but humans are often intuitive. So why don't we see a clear path to create a path, like people have the ability to think. Machines may have simple visual, speech, and other aspects of simulation, but understanding, thinking, creating, and collaboration are very difficult to do . If so, people can never be replaced by machines.
AI development trendThis is an explanation of the future of AI. It is not realistic that we as an entrepreneur do not think how the machine can not be a source of concept. As an entrepreneur, how do we look at the development of today's AI? I have already seen a few clear trends. The first is to implement the AI ​​first strategy in existing products . This year Google proposed artificial intelligence first. Embedding AI technology in all products, such as maps, search, and video, is very reliable, and it has seen very clear benefits, because it has massive users and data. Putting data into the system can greatly increase efficiency. This is what robotics algorithms can do. However, there is a precondition that there must be a product first, and this product has a large number of users. But AI has no such product. We are all building a product using AI technology.
The second is clearer. 99% of the world's AI startups are doing things. I don't make products myself. I open AI to others , such as my cloud identification, computational identification, face recognition, and packaging into APIs. To third parties, many third-party products can open this AI technology themselves, because AI technology has a certain threshold, not every technology can be integrated into the AI ​​team to do this thing. At this time, it is open to others, but as AI companies do not have to do the product itself, this is the vast majority of things that are currently being done.
But its problem is how to make this company continue to grow bigger? At present, the API that is first seen is the user’s willingness and scale of payment habits, which is the problem it faces.
Going out to ask the development modelUnlike many AIs, we have been doing 2C products. There are several ideas in 2C's products. One is the vertical integration of vertical products and the other is the opening of horizontal products . When we develop our own technology, whether it is voice, semantics or search recommendations, we have a set of AI technologies on the market. But we did this technology not for a big and complete scenario, but for the deep integration of our own scenarios. We faced consumers ourselves and we didn't expect third parties to do this AI or turn this AI into a product. it is good. In a specific product, we do a deep integration of vertical products, we do ticwatch, from hardware to software to algorithms are we ourselves do, in-depth optimization around this use scenario. Today, we are very advanced in the world of smart watches, which is not just an AI technology.
If AI is only used in one aspect, such as a smart watch, your value cannot be released. So we have expanded this category horizontally, in addition to smart watches still making products, such as smart mirrors we are about to sell soon, you can imagine your mobile phone with data networking on smart mirror products , You can navigate, listen to music, and drive recorders, including we also do home products.
It seems to be a smart watch, but it cuts into many technologies, whether it is voice interactive smart home products, there are some not shown here, jewelry control and motion sensors, etc. are very core technology in AI. We use smart watches as a vehicle for AI applications. We have embedded many of the AI ​​technologies we have developed ourselves and this set of in-depth integrations. I told a reporter that smart watches are almost one of the most advanced, mature, and concentrated applications of AI today. Many people may think that I am talking, but this is the status of AI. In the future, we think that AI is really going to consume. Scenes, we must do more of this product, and let ordinary people use it and iterate every day to be able to do a good AI, not a very illusionary, in the air that we want to do AI, but there is no specific landing products, this It is a very difficult thing.
Our company has a lot of AI's core algorithms, such as cloud, jewelry, and the accumulation of some techniques of machine learning, but these technologies are very early, such as vision. We finally settled on the AI ​​as the core algorithm, and we did a lot of interactive systems around it. For example, we do TicHome, TicAuto, TicRobot and other products. We will also create a series of products such as smart watches and smart homes. Finally, you can imagine these products. To string together, there will be things that AI didn't think about before. For example, if we use the same set of algorithms at home and in the car, we have a very clear digitization of the same person and we can do a lot of smart things. This is the future of AI startup companies, big companies, and will continue to do this way, through the combination of hardware and software to really make AI, so that AI is no longer on the podium, media, science fiction in the AI It is an AI that you can see every day in your life. This is our vision. thank you all.
Kitchen Hob,Gas Hobs,Gas Cooker Hob,Four Burner Gas Stove
xunda science&technology group co.ltd , https://www.gasstove.be