What is the ambition of Li Kaifu's artificial intelligence and deep learning of Google?

Artificial intelligence is surging in an unprecedented situation. On the one hand, venture capital and entrepreneurial innovation, both regard artificial intelligence as the next treasure that has not yet been reclaimed; on the other hand, application, compared to the stage where the concept prevails, now there is no People's vehicles, AlphaGo, etc. have brought artificial intelligence technology to the situation of "seeing and getting touched".

What exactly is artificial intelligence? What are the elements in this area? How will it change the world today and what problems and bottlenecks will it face? For the application and commercialization of artificial intelligence, which areas will show the fastest results?

In the speech of Tsinghua University's "Tsinghua Academy Computer Science Experimental Class" titled "The Golden Age of Artificial Intelligence", Li Kaifu, Chairman and CEO of Innovation Workshop, gave an in-depth explanation of "artificial intelligence".

In the explanation, the doctor in the field of machine learning traces the source of artificial intelligence, quotes well-known business cases, and combines the layout and adjustment of giants such as Google to outline the framework, elements, commercial fields and conditions of artificial intelligence. The current bottleneck of artificial intelligence and its impact on the current Internet market are highlighted.

It is worth mentioning that the most well-known entrepreneurial instructor in China also advises entrepreneurs who are interested in starting a business in the field of artificial intelligence.

In this long-running speech, the first "scientific style" of artificial intelligence, deep learning, and Google's ambitions was revealed.

The following is a record of Li Kaifu's speech:

thank you all! I am very happy to have this opportunity to come to Tsinghua again, especially under the invitation and introduction of Professor Yao Zhongzhi, my most respected. Professor Yao’s Yao Ban has a good reputation all over the world. From Google to the innovative workshop, I saw that there are many successful engineers who have become the top talents in the computer industry under the training of Mr. Yao.

Before I talk about artificial intelligence, I would like to introduce some of my backgrounds that may not be well known: In fact, before entering several major international companies, that is, more than 30 years ago, I entered the field of artificial intelligence. I was the first natural language processing in 1980, computer vision in 1982, speech recognition in 1983, man-machine game in 1985, VR/AR in 1996... but we now know that at that time I These choices are basically very "bad mistakes" career choices, because every thing, I was involved in the era before its golden age, before the Silver Age, and even when the iron and iron were not. From this matter, in fact, I also want to say that in the field of computer research, the quality of its own is of course very important, but it is necessary to choose the right thing at the right time. I jumped into the field of artificial intelligence too madly at the wrong time. At the same time, the artificial intelligence of the past three or four decades was also ups and downs. It was very hot and fell to the bottom.

But now is the golden age of artificial intelligence. Maybe you will also ask, why is this the golden age of artificial intelligence? To illustrate this problem, this time I definitely not only use some theory to convince everyone, after all, I have done so many "wrong choices" in the past - I also bring some actual data today to share with you why I am today's artificial Intelligence is full of confidence. There are many branches of artificial intelligence, one of which is machine learning. There is also a branch in machine learning that is deep learning. Today I will use deep learning as a case.

Artificial intelligence is a tool

Recently, artificial intelligence has become a hot topic in the world. Many of them are because AlphaGo beat Li Shishi a few months ago. Recently, it was reported on the Internet that it would challenge Ke Jie before the end of the year. But under the heat of this news, one thing makes me feel a pity: everyone's discussion on this topic is focused on whether artificial intelligence is imitating the human brain, whether the "singularity" is about to come, but There is no real concern about the real impact of artificial intelligence on us.

"Singularity" believes that the future machine will have all kinds of intelligence, and humans must do something to protect themselves. None of us here can prove or deny "singularity", but as far as I am concerned, I think that artificial intelligence is a very distant thing to replace people. I think what we need to pay more attention to is that artificial intelligence is a tool that can be used today. It can help humans solve problems, replace repetitive work, and create commercial value. For this reason, I think we are entering the golden age of artificial intelligence today.

Just give a few examples: Most of today's work will disappear afterwards. For example, translation, although it is not perfect, it is very fast every year. In a few years, artificial translation may be very difficult to find. Already working. The same is true for journalists. Today 90% of the Associated Press's articles are written in machines. Almost all jobs where thinking patterns can be rationally calculated will be replaced when there is enough data to support them. Some people say that half of the work will disappear within ten years. Some people say that half of the work will disappear within 15 years. I think these are reasonable speculations.

I think most of you will believe this theory, and if you have doubts about it, you can think about why AlphaGo is so powerful? It is because it can use thousands of machines to play against the tens of thousands of games every day, and this person can't do it; why is automatic driving so powerful in the future? Because it can collect data on the road with its various sensors, this is not a driver can match. So these are some inevitable processes.

What is artificial intelligence?

What is artificial intelligence? I think there may be several parts about it.

The first is perception. Perception includes vision, speech, language; then decision-making. Just do some predictions and make some judgments. These are decision-making levels. Of course, if you want to do a complete system, just like a robot or It is autopilot and it will require a feedback.

What is artificial intelligence?

What is artificial intelligence?

As you can see in these examples, the perception may be more to help identify a baby inside the figure holding a teddy bear on the couch. In the recommendation above, the example I gave is a use of Google now to speculate on what you are going to do through the past. In the bottom example you will see an unmanned car with various sensors. The information it captures can be used to make final decisions, such as how to operate the steering wheel, throttle, brakes, and so on. In fact, the sum of these three things is the artificial intelligence that is summarized today.

Then review the development process of artificial intelligence from four aspects: game, perception decision and feedback. The game doesn't talk too much today, but basically I can see Othello from my college to Checkers and then to DeepBlue chess. After a long period of time, AlphaGo finally defeated the World Championship in Go. We can see from this that this is a journey of more than 30 years.

Milestones in all stages of artificial intelligence development

Milestones in all stages of artificial intelligence development

In terms of perception, from the publication of my doctoral thesis to Nuance becoming a top company, from China, the birth of the University of Science and Technology to the United States, Deep Face, China's Face++, etc., have become better and better, and there have been many in these years. progress. There are some very special examples. For example, I recently saw some funny comparisons. Microsoft Tay started talking to people on Twitter and talked about a bunch of unfortunate words. It was withdrawn by Microsoft, so there are many successful examples here. A lot of interesting events.

Decision-making, from the tools of the early Microsoft Office to the recommendation of Google advertising, and then to the emergence of many intelligent decision-making companies in the financial industry, rapid progress. Google auto mail may not have been seen by everyone, but if you still use gamil, you will find that you sometimes receive emails, Google will jump out and ask if you want to reply, sometimes it will help you write even the reply. And written very accurately. This is also the embodiment of artificial intelligence. Maybe we will not talk in the future, the assistant can help us get it, the assistant of artificial intelligence is definitely a direction.

Finally, feedback, from the early unmanned driving of CMU Boss to Amazon's promotion of logistics with Kiva, to the recent Pepper, Google car, we can see the special heat of this field in the past three or four years, many seem to have commercialized Very good.


Antenk has engineered the d-sub adapters & gender changers that you need for those applications that require equipment that can't be connected directly. 

Antenk Supply a full range of d-sub adapters & gender changers used in a variety of demanding end-use applications. 

Our D-Sub adapters allow you to interface D-Sub connectors with D-Sub or SMA connectors. D-Sub adapters are manufactured in plug to female, plug to male, plug to plug, plug to receptacle, receptacle to female, receptacle to male or receptacle to receptacle gender varieties.

D-Sub cable assemblies are available in every major d-subminiature configuration including DB9, DB15, DB25, DB37, DB50; as well as high-density D sub cable variants like HD15, HD26, HD44, HD62, & HD78.


Antenk's D-Sub Backshells are designed to maximize EMI/RFI shielding for optimal protection and performance. Our comprehensive d-sub backshell product offering features a variety of options that improve the connection stability while maximizing cable strain relief in order to prevent damage to wires and contacts.


The Antenk D-Sub Backshell product offering is available in Die-Cast, Metalized Plastic and Plastic Shell options depending on the application and level of signal protection required against shock and vibration. The Antenk d-sub backshell connector housings are available in up to 5 industry standard shell sizes: 9 pin (HD15), 15 pin (HD26), 25 pin (HD44), 37 pin (HD62) & 50 pin (HD78) positions. If there is a combination of mounting hardware and cable exit angle not available in our standard d-sub backshell product offering above, please contact us directly for any custom or semi-custom backshell design project.

D-Sub Adapters Gender Changer Backshells & Hoods

Low Profile Gender Changer, RJ45 To D-Sub Kits, D-Sub Adapters & Gender Changer,D-Sub Adapter, D-Sub Backshells, D-Sub Hoods,Gender Changers,D-Subminiature Adapter, D-Sub Couplers, D-Sub Splitters, D-Sub Cable

ShenZhen Antenk Electronics Co,Ltd , https://www.antenk.com