Li Yanhong's new internal speech ,2017

The following is the full text of Li Yanhong’s new internal speech:
“Welcome to a new era - content distribution, connectivity services, financial innovation, artificial intelligence”
First of all, it is a new era of content distribution.
Baidu In essence, the core of things to do or distribution of content.
What should we distribute? In the early days, Internet content was based on words. As the bandwidth environment gets better and better, people’s creativity is getting richer and richer, and they have begun to have pictures. To this day, it is actually more than just a picture. There is a new picture-related form called the Atlas which is also very popular and is very suitable for content forms that are spread on the Internet.
In addition to the pictures, there are videos. Today’s video, I think it is very different from the past. The first is the rise of short video, which is very different from the scenes that people reach . We are at the center of content distribution like Baidu, and we must be able to perceive changes in the content distribution environment earlier than others.
Encyclopedia is a very rigorous thing, but it has long been a second to understand video , and it is placed at the top of the entry, so their sensitivity to the environment in which this content is distributed is high enough. So this is a way of thinking that we have to maintain as we embrace the new era. To be sensitive to new things, something will affect you.
The whole world has been changing. From the PC Internet to the mobile Internet, there are many super apps on the mobile Internet, and more and more content is encapsulated in WeChat and Weibo . How to use Baidu’s platform to let content come back, so that our users can easily access it, and complete the mission that we let people get the most convenient and equal information to find the desired? This is something that we need to seriously consider and prepare for and struggle for in the new era.
In the past year, I still feel a little deeper. In the environment of social media and self-media, people often talk about feelings more and want to spread what they are willing to believe. What is the truth? Important. This is a big challenge for a company with a search engine at its core . In this environment, how we deal with it is also a big challenge. Of course, the challenge also means opportunity. I also hope that everyone will be more proactive in thinking about this issue and finding new opportunities that belong to us.
Another content I think is data, and data is new content. After the advent of the mobile Internet era, how can data and applications be combined to present some more valuable things, how can we turn it into new content on our Baidu platform, this is the next step we should focus on.
These structured data, after the completion of these things in the future, on the one hand to provide more valuable content to our users, on the other hand is a prerequisite for our future AI. Because AI’s technology requires a lot of data support, the more valuable and orderly data, the better your final result will be. So that we have a technical big cow to say a word to me : he said that data kills all algorithms.
This has been reflected in many existing products, such as Baidu’s voice search, you can see that the accuracy of recognition has been very high. But when we say, for example, the universal input method to identify these voices, we may not be as good as some other competitors on the market. why? Because we have more data than anyone in this search scene, and this data is very very characteristic and valuable, it is not available to others.
When the user during the search, in fact, he has in mind is clearer requirements, the whole process is a process of looking for answers, expressed from the initial demand, he finally found the answer, especially in the mobile Internet era, in With one hand in such a closed app, we can track every step of his. This gives us enough training data so we can have a high recognition rate. We can not only have a high speech recognition rate, but the whole process allows us to build the best and most abundant knowledge base. We know what it is like from demand to outcome. These things are very valuable to us.
But in other scenes, sometimes even if your technique is particularly good, you may not be able to compete for others. For example, if you take a product to identify it, the result of Taobao’s identification is more accurate than the result of our current photo search. Because a large number of commodity stores are in his head, the merchants have all the data marked. So what is this revelation for us? In addition to the large number of rich queries we have already had to the final user behavior data, we should also actively look outside to find relevant data that are valuable to us. Let them go to Baidu’s platform, we use this data to polish our algorithms and make the best user experience and products and services.
Knowledge maps (including demand maps, user portraits, etc.) are all very basic components of Baidu’s entire artificial intelligence and are our strengths relative to any other company. Moreover, we have attached great importance to it from the very beginning. In the future, we must be more consciously based on our needs, scenes and products, constantly enriching and continually polishing, constantly making this thing to the extreme and achieving the best in the world.
In a sense, our future search engine from index keywords may gradually transition to the engine of index knowledge, which matches the expression of surface text and query, which is the thing of the previous era. The next generation of things is a match between the real needs of the user and the knowledge we have accumulated.
There is also a feed stream product that we value very much now. In the past, the traditional search was that people were looking for information, and now they have to gradually evolve into information to find people. When people do not actively express his information, we can already guess that this is what he likes. This is the information he needs. How to distribute it to those who need and be interested is very convenient and efficient. This is also a new opportunity and challenge in the form of content distribution. Because we have a strong knowledge map and user portraits, we have many years of artificial intelligence technology accumulation, a lot of user behavior and data, so it is very advantageous in this respect.
Second, it is a new era of connected services.
In the next ten to fifteen years, there will be a big trend, which is China’s consumption upgrade. In the past, no matter how bad your product is, you only need it for free, and the user will use it; no matter how good your product is, you will not need to pay for it. Today is not exactly like this. As long as you do well, users are willing to pay for you, virtual things, service things, content things, users are willing to pay. So in our definition, the so-called connection service, as long as it is not purely virtual content, it only contains paid behavior, including online to offline connections, etc., which we all consider to be services. This is also a very new topic for us.
A person from the query to the final service is satisfied, there may be ten steps. We may only do the first step to the third step, or the first step to the fifth step. But people still call connection service, we may not fully control all from end to end, as long as able to push further forward, we will get better user experience one point, we have the ability to control the entire ecosystem will Stronger.
The contentization of services is also a relatively obvious trend we have seen. You write the services provided by glutinous rice into the content that users like, and the mobile phone Baidu pushes the past through the feed , and the user will accept it. What is the user’s favorite content, I summed up the eight words: love and see, pleasing to the eye. You can present the content in this form, and your users will be highly accepted.
What we do is to direct the offline flow from the line. From the form of our products and the stage we are in, this is more suitable for Baidu. The distribution of content and the distribution of services are all about enabling the right people to find the right content and the right services. This is an important issue that we need to find ways to solve the content through the service.
The third era is a new era of financial innovation.
Let’s look at the most important thing FSG does. We call it the B2B2C model.
Internet finance from the initial scene of 1.0 financial, to 2.0 technical finance, I think 3.0 is the so-called model innovation. This is the basic thing that touches the industry. We also believe that for a long time in the future, it is a new era of financial innovation.
We have to find insights. We see what others can’t see, so that we are better than others, and we can do better than others.
Education is a big market. We have advantages over other traditional institutions, but they also have disadvantages. How can we further expand our advantage and make these apps even more super, more sticky, more users, and how to make good use of our big platform. The industry of finance can make something innovative.
The whole financial can be covered in a wide range, and FSG has done a lot. I won’t elaborate on it, but we feel that such an era is also a new era of financial innovation.
Finally, it is a new era of artificial intelligence.
An obvious opportunity is the combination of hardware and software. Amazon ‘s echo is very successful, and the logic behind it is that it has made some modifications on this hardware, not just a software solution. Mobile phones should consider the issue of power consumption. But smart speakers don’t need to worry about power consumption, which makes it ready to respond to your needs at any time . So this is the opportunity for new artificial intelligence brought about by new hardware forms.
The unmanned vehicle integrates the radar and camera hardware, which makes it capable of driving automatically under unmanned control. It is not a pure software requirement. And this kind of innovation of this kind of integration, in fact, is now at a very fast speed in the iterative process.
This time we released a new product at CES that is equipped with a secret fish at home, which is also a typical example of hardware and software combination. These abilities are not just about the ability of your face recognition to recognize him, but to say that you have a piece of hardware where you can monitor your home environment at any time. Therefore, this integrated innovation of software and hardware will have many opportunities in the future.
The second opportunity is the movement of the industry. Artificial intelligence changes every industry. For every industry, you only have to seriously think about it, you will find that there are very many opportunities for application.
We are not a vertical industry company, we are more like a platform company, we can not very very understand every industry.
But we are close to some industries, such as the medical industry. At first, Baidu’s medical treatment was very simple. It was an O2O thing. We have users, many people come to see where to see a doctor. How can we help him hang up the number he wants to hang? But since last year, we have found that in fact, intelligent consultation has become more and more practical. If our intelligent consultation system can reach the average level of a doctor’s occupation, then it is entirely possible to pass an intelligent system, at least to assist these doctors to make some judgments. It is really a time to go to the hospital when the illness is serious. Go to a more large-scale hospital.
Then go one step further, like gene sequencing, which is a very big data item. If you can combine medical knowledge with artificial intelligence, and find the answer, it is also a big breakthrough. Going down, for example, the development of new drugs, how to use artificial intelligence technology to find new drugs, these things are still very very early, but we think that in the end, these things can make a difference. I also believe that in logistics, in retail, in many, many areas, there is actually a similar solution that uses artificial intelligence to help solve.
The application of artificial intelligence at the enterprise level is also very extensive. An example was mentioned in Baidu World last year, called Gold Sales. We identify the customer’s problem in real time, and then promptly tell the salesperson how the most sales of the customer answer the customer’s question. At present, we already have a team working on this project, which is the real advantage of using Baidu’s artificial intelligence. Such a project that can generate value for the business. There is also this intelligent customer service, most of the problems are repeated, you can use the machine learning method to find the answer.
I think there are still a lot of similar applications, but we don’t have the energy to go one by one. But we need to have such a mentality to think. In the era of artificial intelligence, what your thinking logic and way of thinking should look like is not very clear. But I believe that it is unclear to us and less clear to other companies. We should do our thinking and try in this direction as early as possible.
Meeting the new era - we must be prepared to fight every war, battle and battle
Most importantly, the company has a good atmosphere. In the past year, everyone felt that the intensity of anti-corruption is relatively large, which is consistent with this concept. On this battlefield, we must always be persistent. If we want everyone to feel fair, we must fight those who have lost the loopholes in the system. We are to go towards a more positive direction, the company will become a fairer, more equitable for everyone a company.
Second, the user is supreme. We need to think more about the needs of the market and think more about the value of our products to our users.
When we go down and disassemble the task, it is easy to become a bad experience for some users. This situation is actually very dangerous. If you use a simple mechanical KPI to break down and pass, then it is very likely that when the grassroots level to the frontline staff, he does not understand why we should do this. At this time, if his mind is not about the user experience, not thinking about the market demand, thinking about how I can complete this number, then the company is really a long time to finish. What employees do is not what we want to do.
So we need to think about the problem from the perspective of the user. Although we are now a native app, there is no real native app experience in the process of using it. We often click on a search result and go to someone else’s website. Traditionally, there are a lot of ads on other people’s websites, and that’s what other people’s websites are doing. But from the user’s point of view, users always think that he is using Baidu, we are responsible for the user’s needs.
The third one is that you can’t talk about seniority. Starting from real needs, and based on human capabilities, especially in a rapidly changing industry like the Internet, many young people actually feel this change earlier than we can. To allow them to have enough room to play, it is necessary to use new people boldly.
Ten years ago, when Baidu made some new products, the person in charge of those products might be ten years younger than the people we are here today. But the person present here, you will easily take a person who is ten years younger than you, this product is handed over to you? You may not be relieved now. In fact, it doesn’t matter, they may be better than you.
We also want to encourage open communication. Tolerate these flamboyant personalities, Baidu culture includes me, and more is to do things relatively mild. But in fact, those who have horns and horns, those who speak more fluently, those who have some “smelly faults”, we want them to have opportunities on the big platform.
Also encourage people who drive change. We have been doing it for more than a decade, and many things have already had inertia. I don’t think about whether there is a better way.
If you are in the team, the person next to you, you think he is very poor, you are ashamed to be with him, how can this team have combat power? So lowering the requirements is like the shortest board in the barrel, and the water flows away from there.
To meet the new era, we must fight every war, battle and battle.
The whole company is playing a war. From a holistic perspective, artificial intelligence is the foundation of a technology for the entire company. The data for this precipitation must be precipitated. The integrated technology should be integrated in one place, and the openness should be open to all business units. This is a game of chess.
Content distribution is our core. We can exist. The reason why we can do a lot of business is because we have a strong market for content distribution. We must understand this principle.
Our approach must be based on user value, taking into account diversified development and ecological self-construction. For many years we have been talking about focusing and doing one thing. Since last year, we have to consider more than one business. At the same time, we must build our own ecology. The original ecology is gone. We need to build our own content ecology, our service ecology, and our financial ecology. This is a unified thing for the whole company.
The battle is every product. Every PGM is commanding a battle. Whether you can win or not depends to a large extent on your personal ability and your leadership.
Some of the smaller problems or one project is fighting. Fighting is also very important. Our morale is often driven by snoring and is inspired by winning these battles.
You can be beaten with morale. You can do it every day, and you will do it step by step. This morale will never be good. IDL’s strongest brain project, which was often discussed at 3 or 4 o’clock, was sometimes excited to cry. Because the strongest brain is a battle, you are very excited to win, and the whole team morale can last for a long time.
Can win, this leader can be promoted. It’s not how long you have been waiting, it’s not that you have a better offer outside, you can get better treatment here – you won, you will be recognized.
To meet the new era, we must also clean the door.
Combating false information and excessive advertising is the most important factor that currently hurts our experience. As one of the core platforms for our content distribution , users are considered to be using Baidu regardless of whether they jump twice or three times. So you have to give him the same quality experience as Baidu. What do users think, you have to recognize. You can’t say, I want to educate him that this is not Baidu, you can’t educate.
Second, PC search has a lot to do with moving. After the PC searched in the browser today, you went out to another website, and the advertisements were flying all over the sky. In fact, we have the ability to control these things to a large extent.
Then, we must eliminate products that are not competitive in the market. If you can’t do it, don’t mix it there, don’t stay there. The withdrawal will be withdrawn, the customs will be closed, and the merger will be completed. Resources focus on projects that we have advantages and strategically important.
Management also needs to have metabolism. If you can, you can’t. Good is good, bad is bad. If you do a good job and you don’t do well, you should take measures.
These things are done, we can be ready to meet a new era.
Such a new era may have threats, challenges and opportunities. This is a future we have to face.
We have to prepare for it, we have to find opportunities.
I am also very confident that in such a brave new world, we will do better and better, thank you!