Baidu He Junjie: AIGC technology is becoming a new variable of Baidu Mobile Ecology

AIGC will set off a revolution of content creation. Will this be the gospel of "fighting workers"?

The debate about AI creation seems to have never stopped.

Recently, an AI painting award -winning incident quickly broke the circle, allowing a large number of illustrators to send out the emotion of "unemployment": In August, in a art competition in the Colorado expo in the United States, a game designer used AI painting tools Midjourney to create A portrait of "Space Opera" and the first place in the art competition.

This painting has spread between the art circle and the majority of netizens, and controversy is constant. Critics are angry and think that AI plagiarizes a large number of predecessors' works. Its creation has no emotional and soul at all, and cannot be compared with human art creation.

Of course, AI paintings are also a fans: even if the creator modifys the text description to make AI paintings over and over again, the work is still quite ornamental, and AI's creation has unique value.

The former emotions are easy to understand: once in the context of "replacement of AI", workers can easily cause self -worth doubt and panic. But even so, it is difficult to kill the AI ​​painting technology that is still in the 襁褓, and the trend of the AI-GENERATEDCONTENT (AIGC) behind it.

Domestic Internet company Baidu has also launched the AI ​​painting generator "Wenxin · One", which can be used in Chinese to generate images.

Obviously, AIGC technology has become the emerging torrent of the digital age.

At the 20.22MP · Baidu Mobile Ecological Conference held on September 23, Baidu released AI assistants, which can be called AI workers' group, covering various AIGC applications, including AI automatic text, pictures, and pictures converted into videos. In the words of Baidu, there are not only copywriting, illustrator, but also video producers in this team. You can hand it over to AI for staying up late. The difference said: "Only the world where AI workers were injured."

In addition to the technology Hewlett -Packard, Baidu seems to have more expectations for this migrant group. In May of this year, He Junjie was officially promoted to the senior vice president of Baidu Group as the general manager of the mobile ecological business group (MEG). On the eve of the Baidu Vientiane Conference this year, he said in an interview with the media for the first time that the user spent more and more time on short videos, and the video content will become more and more important. Baidu will be hugged as always.

The picture shows He Junjie, the general manager of Baidu's senior vice president and the general manager of Baidu Mobile Ecological Business Group (MEG)

At present, the short video campaign is becoming more and more hot. If Baidu wants to launch a shock again to the short video and kills the siege, it needs a more lethal secret weapon. AIGC is regarded by Baidu as a sharp blade, and an attempt to compete with the current video platform that comes to the current video platform.

With the power of AIGC, Baidu hopes to set off a content production revolution. In other words, it brings a set of AI production content tools for Baidu's existing more than 5 million creators to help them get video content faster and better. Head short video platform.

At the conference, He Junjie, the person in charge of Baidu Mobile Ecology, believes that AIGC technology will become a new variable and new direction for Baidu Mobile Ecology. Based on AI production copywriting, pictures, graphic transfer video and other technologies, Baidu has established the "AIGC Media Alliance" with dozens of authoritative media.

At present, the mobile ecological Internet has seen the top. How to wake up the sleeping traffic is a problem facing each family. Searching for old cannon Baidu is no exception. And when Baidu AIGC's sharp blade came out of the sheath, we asked, is this sharp blade sharp enough? Can Baidu mobile ecosystem be harvested as scheduled?

AIGC, the gospel of the worker

In recent years, AIGC (AI production content) has been regarded as a new type of content creation method after PGC (Professional-GenerableContent professional production content), UGC (User-GenerableContent user production content).

In the past, PGC mainly relied on professional team creation, just like movies, TV series, and game production. The cost is high and limited. It is difficult to operate without a special team. The short video platforms such as blogs and today's Douyin Kuaishou are typical UGC cases.

At present, the timeliness of users' timeliness of content production is getting faster. Many short video users are "don't panic in the event of trouble, take out their mobile phones and shoot", "alone music is not as good as public music" has become the daily life of the masses. However, the efficiency of PGC and UGC output content can no longer meet user needs. At the same time, there are a lot of boring and duplicate work in the content production process: for example, the migrant worker stays up late to find a suitable GIF animation material.

The emergence of AIGC is expected to assume part of the boring work for the worker.

In 2021, GARTNER, a global authoritative IT research and consulting company, released 12 important strategic technical trends in 2022, among which the generating artificial intelligence (General Intelligence) was placed at the first strategic first place Essence Gartner said that the generation AI can learn from existing data (image, text, etc.) and generate new and similar original data, which can be used in the fields of automatic programming, drug development, visual art, social networking, business services and other fields. Although the range of generation AI coverage under Gartner definition is wider than the AIGC discussed earlier, the trend of AI generating content cannot be ignored. Gartner predicts that by 2025, generating artificial intelligence will account for 10%of all generated data, and the current proportion will be less than 1%.

At present, AIGC is mainly used in content, entertainment and other fields. For example, in the field of content, most of AIGC appears on the platform to automatically generate videos and pictures for users, which greatly improves content production efficiency. Xinhua News Agency's "Quick Pen Xiaoxin" (Sports Finance's News Automatic News) and AI painting are all this category.

Especially for AI painting, some world-renowned companies have also joined the battlefield, including the text generation image tool of the artificial intelligence research company OPEN AI, Dall E; Google's AI drawing tool Imagen; Meta AI developed Make-A-Scene; Microsoft and Beijing The AI ​​drawing tool Nüwa, which is developed by the university cooperation with "Nuwa".

In the field of entertainment, AIGC often uses virtual digital people as the core carrier, becoming one of the core basic technologies of the concepts of the Yuan universe and digital people. For example, Li Weiki, a virtual number person with bytes beating, Alibaba's virtual employee "Ayayi" is a case of AIGC empowering digital people.

Baidu is caught in content and entertainment.

From the perspective of the AIGC application released by the Vientiane Conference, Baidu's AI Copywriting Assistant, AI Illustration Assistant, AI Video Production Assistant and other AIGC applications are multi -mode content understanding and automatic generation technology, mainly for extensive users and content creators.

According to Baidu He Junjie, the AI ​​copy assistant can quickly produce different styles of articles and content paragraphs according to the needs of the creators.

The AI ​​illustration assistant is based on Baidu's heart. According to the keywords, it can automatically "draw" more than ten different styles of paintings, so that the creator can achieve "free pictures".

AI video production assistant is based on the natural language processing and cross -modal AI technology of the large -scale model, which automatically generates the video of the creator's graphic, and the speed is "minute -level". above.

Earlier, the National Early Warning Information Center was based on Baidu Baijia's AIGC graphic transfer video technology, shortening the time to produce an early warning information video to 90 seconds. Within two years, the project has achieved a total number of national warning matrix 1 million, with a number of touches of 40.8 billion.

Another case is that after the use of the Baidu Baijia AIGC graphic transition video technology, the production time of a news video can be reduced by half an hour after using the Baidu Baijia AIGC graphic transfer video technology. However, in the past, the editor should "retrieve material, integrate, pull the film" to form a basic version, and then conduct step -by -step review and release. Now editing is the basic content of pre -processing.

It is worth mentioning that the broadcast volume of graphic transitions is nearly 10 times more than the same graphic. This also means that the picture -and -text transfer video processed by AI has not lost the original creation of users in terms of effect presentation.

It is understood that Baidu's "Creator AI Assistant Group" has begun to invite internal testing and will be opened to all creators from the fourth quarter of this year. Under the efficiency of efficient content production, AIGC technology is setting off a content production revolution.

At the same time, AIGC often appears as the core ability of digital people, and is accompanied by digital people. Similarly, Baidu also "three -in -one" of digital human technology and AIGC graphic transition video technology, TTS voice synthesis technology, and customized real -life twin digitals for the media and creators.

It is understood that Baijia No. Based on real twin digital technology technology, the official short video brand "Green Beeman" of the China Youth Daily customized the digital host "Qing Xiaoxia". You only need to enter the text, "Qing Xiaoxia" can automatically complete the broadcast. If it is not prompted, it is almost difficult to find that it is not a real person.

In addition, "Qing Xiaoxia" can also be competent for multi -language broadcasting, sign language hosting, 24 -hour automation broadcasting and other tasks, and contracted all the reporting reports with their own power. Live reporters can bring more having more having a diligent digital man behind the tireless digital people. Reports of unique value and viewpoints.

Baidu He Junjie also said in an interview: "People are limited. If you put your energy on repeated mechanical labor such as editing videos and broadcast news, then the time on thinking about the depth of thinking and the innovative point of view will be correspondingly corresponding. Reduce. Now with the corresponding AIGC technology, creators can consider more high -quality topics, deeper insights and more likely content. "

Therefore, it is difficult to say that the AIGC revolution is to completely replace the existing PGC and UGC. As an efficiency tool, AIGC has brought acceleration of content production. It still requires human control and calibration in creative and artistic aesthetics, and human creative value still cannot be replaced. But in terms of boring and invalid internal volumes, AIGC has its existence. Ten years of investment with Baidu's new facade

The emergence of multiple AIGC applications on Baidu is related to its continuous evolution of AI technology models in recent years.

In the past, AI training often relied on specific data models. For example, AI writing poems is based on model training on the basis of massive poems data sets, imitating human writing thinking, and automatically generating verses. The same is true for AI painting.

In recent years, AI capabilities have been continuously improved. Writing poetry is closer to human writing, and the rendering images have also been more fine, but specific data training sets are only suitable for targeted scenarios and poor universality. For example, an AI trained by a model will only write poems and cannot create news, videos, etc.

Therefore, the development of artificial intelligence has begun to turn to a pre -training model with high generalization capabilities and universality. AI technology can be briefly migrated on the basis of the same model to adapt to different work scenarios.

Taking Baidu's AIGC as an example, the technology is also inseparable from the pre -training large models, deep learning platforms, and artificial intelligence chips. Since 2019, Baidu's deep cultivation training model has been developed, creating a family of flying paddle -hearted models.

It is understood that the Wenxin series model has the core feature of "knowledge enhancement". Based on the semantic understanding framework of continuous learning, it is integrated from large -scale knowledge and massive data. For example, AI painting ability uses a large model of Wenxin-cross-modal graphic generating model Eric-Vilg; AI digital person dialogue ability, using dialogue to generate big model-Wenxin Plato. At present, Baidu Wenxin has released more than 10 large models, which greatly reduces the threshold for AI development and applications.

The investment from Baidu can also be glimpsed. According to Baidu, the company has invested 10 years in the AI ​​industry, with a cumulative investment in R & D billion, and has ranked first in China in terms of AI patent applications and authorization volume for 4 consecutive years. This may be the reasons why Baidu can bring AI copy assistants, AI illustration assistants, AI video production assistants and other AI workers at the Vientiane Conference.

But objectively, AIGC's completely mature tense has not yet arrived. Many netizens also found that when using AI painting, although the portrait of AI is also very fine, sometimes there are still random patchwork, and the painting style is obviously plagiarized.

Where will AIGC go in the future? Earlier, Li Yanhong, the founder and chairman of Baidu, judged that AIGC will go through three development stages in the future: the assistant stage, the collaboration stage, and the original stage.

In the first stage, AIGC will assist humans to produce content; in the second stage, AIGC will appear in a virtual human form that coexist in virtual and real, forming a situation of human -machine symbiosis; the third stage is the original stage, and AIGC will independently complete content creation.

Baidu also sets a FLAG: In the next ten years, AIGC will subvert the existing content production model. It can achieve an original content of AI at a cost of one -tenth of the cost of one -tenth of the cost. When netizens are surfing online, they may see more content produced by AI.

In other words, Baidu will also invest in the AIGC field for ten years or even longer. In retrospect, the previous cutting -edge technology that allows Baidu to persist in R & D and investment for ten years is autonomous driving. It can be seen that within Baidu, AIGC has been mentioned at the level of strategic level.

This may be related to Baidu's current external environment. Even the Internet giants, which started with short videos, began to expand the search business based on the original business, trying to use "video+search" to eat traditional search+information flow models. Searching for Baidu also needs to keep up with the trend of short videos.

AIGC can be understood as a attack by Baidu's mobile ecological business in the field of short video. "The best way to seize the trend is to make your advantages grow on the trend." Baidu He Junjie said at the conference that the mobile ecology is the "first stop" of Baidu's artificial intelligence technology. Can come out.

Baidu AIGC applications have aimed at the convenience and efficiency of content creation. When users and creators find that AIGC can save themselves half of their creative time, the flow of traffic is imaginable.

In addition, Baidu also intends to allow creators to have a digital person "avatar" per capita, which should also attract many content creation groups. In addition, after the generation of AIGC content, the creator naturally faces the huge traffic pool of Baidu Baijia. From content creation, distribution, to exposure support, Baidu has a series of new gameplay.

Last year, Baidu's mobile ecological business shouted the slogan of "personalization and service" at the Vientiane Conference, trying to break the single information search island in the past and make Baidu a platform to provide "information+service".

The new AI technologies and applications announced at the Vientiane Conference this year can be regarded as the continuation and new variables of Baidu's strategy, so that its "information+service" strategy is more thorough. More importantly, the AIGC air outlet is coming, and Baidu's first layout is expected to become a new tentacle to break the mobile Internet traffic.

Just as every new thing is facing doubts, AIGC has a lot of challenges.However, from the many content revolutions experienced in history, it is difficult for new content production methods to completely replace old artistic creations, just like photography to paint and TV.New technology will establish new order and ecology, but each art will have unique audience and existence value.AIGC is the product of the digital age.One day, AIGC will also become ordinary, becoming the same content production method of UGC and PGC.Just like in the two -dimensional community, users are watching the release of AIGC every day, or urged in the barrage comments.This will be a feast of AI technology and humanities.

- END -