Baidu is on the trend again! Why are technologies like ChatGPT so popular among car companies?

At the beginning of 2023, the chatbot ChatGPT suddenly became a phenomenon. This new type of AI software can not only chat and communicate, but can also write essays and poems, and can even make up various stories and popularize various physics, mathematics, and economic principles.

The emergence of ChatGPT feels like the emergence of AlphaGo a few years ago. At that time, it was also the focus of the whole people. In the end, the entire Go world has undergone earth-shaking changes.

Almost at the same time, the Chinese version of ChatGPT Baidu "Wen Xin Yi Yan" was also officially launched, and a number of media companies announced that they had become ecological partners of "Wen Xin Yi Yan". More importantly, car companies have also begun to enter the market one after another. Geely, AIWAYS, Lantu, Great Wall, etc. have announced that they will connect to Wenxinyiyan.

Why can a chat software arouse such high popularity? Why are car companies so interested in Wen Xinyiyan? Let Kung Fu Motors take everyone to see it together.

(1) The core is content generation, does Tom Cat and Baidu know about it?

First of all, we need to understand why this time ChatGPT, or Wen Xinyiyan, can bring such a big shock to the market. The core reason is that they are very different from previous chat or search software.

All ChatGPT-like software can be regarded as derivatives of ALGC technology. The so-called AIGC refers to AI Generated Content, which uses artificial intelligence technology to generate content.

To put it bluntly, chatting in previous software was just chatting, such as the earliest Tom Cat, and the recently popular Xiaoai classmates or Xiaodu. Although they have certain in-depth semantic understanding or dialogue capabilities, they have obvious deficiencies in knowledge reserves and computing power, so they cannot really solve the problem.

Although previous content search or question-and-answer software, such as Google, Baidu, Zhihu, etc., could provide certain answer information, these questions and answers mostly came from thousands of netizens. This leads to a problem. When we search for a question, it is difficult to identify which answer is correct in most cases.

ALGC technology is based on a massive database and a new AI framework, and uses its own special algorithm model to give the most reasonable answer. Some netizens asked Wen Xinyiyan to do math questions for the high school entrance examination, and the accuracy rate could reach more than 80%. This is a very scary thing. In other words, Wen Xinyiyan may already be "smarter" than most ordinary people.

If the previous chat software can only relieve boredom, the search software can only be used for retrieval, and the final main operation and judgment group is still humans. Starting from ChatGPT and Wen Xinyiyan, chat tools will move towards efficiency tools. Although you can only do high school entrance examination questions now, as the software continues to evolve, in the future you may do research and solve large-scale modeling and calculations.

Kung Fu Motors has a friend who works at a car company, and he expressed great concern. Because his main job is to do simulation calculations of the power and economy of cars, he said that once Wen Xinyiyan evolves to the point where he can model his own models, their entire industry will be wiped out. Maybe ordinary high school students can do it. The bar will become very low.

Of course, we are still very far away from that distance, but it seems that Pandora's box has been opened.

(2) Car companies are flocking to them, are they afraid of missing out?

According to past experience, it often takes a long time for a new technology to be applied to a mature field like the automotive industry. Smartphones began to rise around 2010, but until 2019, Carplay was still the best among mainstream car systems. It was not until the rise of Tesla and a number of new forces that this phenomenon was ended.

So why did so many car companies announce their access to it as soon as Wen Xinyiyan appeared this time? In fact, it is still because car companies are afraid. Because of this hesitation, an era may be missed.

Nowadays, in-vehicle voice has become a very popular method of in-car interaction. The most typical one is Wenjie. With the excellent interaction capabilities of the Hongmeng system, it has even given this brand an aura of intelligence.

However, shortcomings of in-vehicle voice still exist, and the biggest difficulty is "semantic understanding". Because there is no intelligence in "understanding" voice commands, everyone always needs to use a specific voice and wake up specific functions. This is essentially "assembly language" rather than "natural dialogue." The emergence of Wen Xin Yi Yan allows people and cars to achieve more efficient and flexible chat and communication, and everyone can give instructions in a more relaxed situation.

Of course, Wen Xinyiyan’s current “answer” is still not smart enough, and it needs to be optimized based on application scenarios. This is what car companies need to do after accessing the system.

Another major advantage of Wen Xinyiyan should be its use in intelligent driving, or future driverless driving. In the past few years, Baidu has begun to use AI technology to provide standardized solutions for the industry. Its intelligent transportation solutions have been implemented in nearly 70 cities. Among them, the implementation of "unmanned" is not low. Baidu Baidu has provided fully unmanned travel services in Wuhan and Chongqing, and is a leader in this regard.

Car companies' access to Wenxinyiyan is also intended to be a port for Baidu's intelligent driving.

(3) After a good start, how will Baidu play its cards?

Judging from the current popularity and technical reserves, the domestic AI market is about to experience explosive demand growth, and the commercial value it will bring will be extremely exaggerated.

As a leader in this field, Baidu has stood on the top of the wave, as evidenced by the popularity of Wen Xinyiyan.

In the era of AI artificial intelligence, the biggest change is actually the underlying logic of the technical architecture. The previous technical architecture was actually divided into three layers: the chip layer, the operating system layer and the application layer. The core of which was to utilize the computing power of the chip.

However, with the development of technology, computing power has begun to become excessive, and the technical architecture has begun to be divided into four layers: chip layer, framework layer, model layer and application layer. Deep learning frameworks and pre-trained large models are becoming more and more important, and they are even the core of the entire technical architecture.

In this regard, Baidu, as a full-stack artificial intelligence company, does have innate advantages, which is why car companies have sought cooperation with it.

But this lead is actually not stable, because technology iterations are often sudden. For example, in the field of pure electric vehicles, traditional car companies have not made breakthroughs for decades. Tesla went from the niche "toy" Model S to the popular Model 3 in only four years. Another example is hybrid. In the past ten years, car companies have only produced about 200,000 vehicles a year. BYD DM-i has achieved nearly one million vehicles in two years after its emergence.

Now Baidu does have a first-mover advantage, but how to play this card is a problem. If we were to search for coverage alone, it would probably be like Carplay. Because it has to cover more devices and its functions can only be simplified, it will be easily surpassed by latecomers.

However, if we combine it with Baidu Apollo and do deep binding, it may cause resentment from car companies. After all, no one wants to hand over crucial intelligent driving to other companies. What's more, Baidu has stopped building cars and launched Jidu, which will become a thorn in the hearts of car companies.

How to choose a method that is acceptable to car companies, yet has wide enough coverage and a profitable business model, will be the biggest problem facing Baidu.

(4) Kung Fu shooting

In fact, the main focus of Chinese technology is "accumulation and development". Baidu quickly came up with Wen Xinyan to respond to ChatGPT because this technology has long been mature and has actually been used in many fields before. After competing software appeared, it was quickly launched based on its characteristics and immediately aroused a great response.

Car companies are vying to carry it, and what they value is Baidu's technology accumulation in the AI ​​market and its research in the fields of smart cockpits and smart driving. In this wave, Baidu is on the forefront.

But the problem still exists, that is, how Baidu plays Wenxinyiyan's card well. Playing well may be a good thing in the future, but if you are not careful, you may miss an opportunity.