At 18:18 on April 16, Liu Qiangdong's AI digital human made his live broadcast debut on Jingdong App. In less than 1 hour, the live broadcast room had more than 20 million views. It is reported that the image of this AI digital human of "Procurement and Sales of Dongge" was created by JD Yunyanxi, and it was also unveiled in the live broadcast room of JD Home Appliances and Jingdong Supermarket.
Netizens who watched the live broadcast were amazed that this "Dongge" was very similar to Liu Qiangdong himself, and the small movements, mantras, Suqian Mandarin and other designs almost 100% restored Dongge's expression, posture, gestures, and timbre.
Dongge digital people keep bringing goods, during which they don't chat, and they don't need to drink water. Complete the live broadcast process such as product introduction + getting on the car in one go, and it only takes 20 seconds to complete a series of operations.
And netizens are also watching the emergence of the digital "procurement and marketing Dongge" with relish. According to the disclosure of Jingdong Supermarket, the number of viewers in the live broadcast room exceeded 10 million in just 40 minutes, creating the highest peak in the number of viewers since the launch of the live broadcast room of Jingdong Supermarket, and the number of viewers exceeded 20 million in the past 1 hour, and the average stay time of users during the live broadcast period reached 5.6 times the daily average. Within 40 minutes, the overall order volume of the live broadcast room exceeded 100,000.
"Break the kettle" and spell out your own "traffic password"
In the era of live broadcast in the ascendant and AI large language models are booming, JD.com, as a domestic self-operated e-commerce company, must break through the traditional e-commerce model and cut into the live broadcast e-commerce track. And Douyin, Kuaishou, Xiaohongshu and other platforms have become more and more mature live broadcast ecology under the "pinch", Jingdong must "break the kettle", at this time, with virtual human live e-commerce as the entry point, it may be an opportunity for JD to surpass Douyin, Kuaishou and other live broadcast tracks.
So for JD.com, this "virtual person" is none other than the founder Liu Qiangdong. Because to a certain extent, for the public's perception, Liu Qiangdong is equal to Jingdong.
Of course, Jingdong is also "the drunkard's intention is not to drink", in addition to creating a virtual image and paving the way for the brand to enter the live broadcast e-commerce, this JD Yunyanxi voice model itself is very competitive. During training, 50,000 hours of massive and fresh voice data were "fed", which allowed Yanxi digital humans to intelligently match different live broadcast styles, such as creating a professional atmosphere with calm timbre, or attracting users to place orders with a very infectious voice, and also giving Yanda a physical performance, which can start broadcasting in as fast as two hours and can be broadcast live 24 hours a day. Experiments show that the vast majority of users are difficult to perceive within 120 seconds, this is a digital person, compared with real anchors, the cost of JD Yanxi digital human is less than 1/10, and at present, JD Cloud Yanxi digital human has served more than 4,000 brands.
The market size of the digital human-driven industry is expected to exceed 600 billion
In addition to JD.com, Baidu, TikTok, etc. have also seen the opportunity of digital human live broadcasting. Baidu e-commerce has launched its self-developed "AI full-stack digital human live broadcast solution" Huiboxing. TikTok is also developing an AI-based virtual digital human feature to better help merchants on e-commerce platforms promote and sell.
Also on April 16, Baidu founder Robin Li was once again active in the public eye, standing for the tool version of Wenxin Model 4.0. Wenxin Yiyan's users have exceeded 200 million.
Robin Li said that large language models themselves do not directly create value, and AI applications developed based on large models can meet the real market demand. ”
According to iiMedia Consulting data, the market size and core market size of China's virtual human-driven industry will be 186.61 billion yuan and 12.08 billion yuan respectively in 2022, and it is expected to reach 640.27 billion yuan and 48.06 billion yuan in 2025, respectively.
With the release of ChatGPT and "Wenxin Yiyan", the intelligence level of virtual humans is expected to be greatly improved, and it will be applied in different fields, such as live streaming, which can help achieve 24-hour barrier-free communication with customers.
Can AI completely replace live streaming?
Although Liu Qiangdong's digital person is quite similar to a real person, whether it is his appearance or voice tone, he imitates Liu Qiangdong's habits, but many netizens still commented that it was too blunt, and many netizens shouted to Jingdong and asked Liu Qiangdong himself to broadcast in person, "The copycat version of Dongge still lacks some feelings" and "Like customer service, he will always be a robot"......
In this era of rapid development, consumers seem to be looking for some sincere, emotional parts in the online world, whether it is Dong Yuhui who exported a chapter in the live broadcast room a while ago, or Luo Yonghao, who "makes a friend" and sells cross talk, people pay more attention to the real and vivid characteristics of being a "real person".
In addition to the choice of people, the current platform specifications are also being improved. Previously, Douyin had released the "Douyin Platform Specification and Industry Initiative on Artificial Intelligence-Generated Content", mentioning that the image and content of virtual anchors must be pre-recorded, and virtual anchors who have not been reported by the platform cannot start broadcasting. In addition, the virtual live broadcast must be prominently marked to help users distinguish between virtual and reality. The platform will provide a unified AI-generated content identification capability to help creators mark and make it easier for users to distinguish between virtual and reality.
(Part of the source: Jimu News, chatterbox)