Generative AI management rules are implemented, and the post-big model era is coming

The "Interim Measures for the Management of Generative Artificial Intelligence Services" jointly promulgated by the State Cyberspace Administration of China and other seven departments has established a development standard for China's generative AI industry.

The purpose of this regulatory document is not limited to control, but also uses a lot of pen and ink to convey the principles of promoting industrial development, especially emphasizing "encouraging the innovative application of generative artificial intelligence technology in various industries and fields."

The timing of the release of the new regulatory policy coincides with the cooling down period of the AIGC boom. At this time, the appearance of various large-scale models has been unable to bring new sensory stimulation to the world. Although the upgrading of large-scale models is still a long way off, the outside world's desire for applications has prompted the era of large-scale models to come early.

Kaiyuan Securities Research believes that games, e-commerce and education will be the first tracks to be transformed by generative AI. When the large model brings a subversive experience in subdivided scenes, AI, whose momentum has weakened, will regain its charm.

Generative AI management rules will raise the concept stocks

On July 13, the State Cyberspace Administration, together with the National Development and Reform Commission, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration of Radio, Film and Television, announced the "Interim Measures for the Administration of Generative Artificial Intelligence Services" (hereinafter referred to as the "Measures") ), which will come into force on August 15, 2023.

After more than half a year of chaotic running of generative AI at home and abroad, China has made clear management regulations on the research and development and services of generative AI, marking the end of the AI industry's reckless era.

Netcom Office publishes generative AI management files

This regulatory document can summarize two key words: strengthening supervision and encouraging development.

At the regulatory level, the "Measures" point out that as long as generative AI technology is used to provide services such as text, pictures, audio and video to the public in China, they need to abide by the regulations. In other words, overseas generative AI applications that provide services to domestic users are also included in the scope of supervision.

Regulatory regulations have made detailed regulations on the research and development and service links of generative AI.

In terms of content, content that is prohibited by laws and regulations such as endangering the country, violence, pornography, and falsehood cannot be generated; at the level of algorithm design, effective measures must be taken to avoid discrimination such as ethnicity, belief, and gender; in terms of rights and interests, intellectual property rights must be respected, and the portrait rights of others must not be infringed , privacy rights and other rights and interests; in terms of service specifications, providers should fulfill network information security obligations, take measures to prevent minors from indulging in generative AI services, and must not illegally collect users' personal identity data.

The above regulations have basically drawn a red line for the negative behaviors exposed by generative AI in the past. For example, vicious events such as using ChatGPT to generate rumors, infringing intellectual property rights, and using image-generating tools to "change faces" and "undress" will be strictly regulated in the future.

In addition to necessary supervision, promoting the benign development of generative AI is also a key point emphasized in the "Measures". The words "encourage" and "promote" appear in the document several times.

The regulatory authorities stated that they will encourage the application of generative AI in all walks of life, encourage the independent innovation of basic technologies such as algorithms, frameworks, and chips, and promote the construction of public training data resource platforms. At the same time, it is necessary to promote the sharing of computing power resources and improve the utilization efficiency of computing power.

Under the principle of emphasizing supervision and promoting development, China's generative AI has entered a new stage of development. On the one hand, after the regulatory regulations are clarified, relevant companies can implement research and development more clearly and efficiently, and accelerate technology implementation; on the other hand, the multiple incentive regulations in the new regulations can also reduce the cost and increase efficiency of generative AI research and development.

Globally, the European Union is advancing the final legislative process of its "Artificial Intelligence Act", the United States is expected to introduce a series of regulatory measures this summer, and Japan is expected to introduce regulatory policies before the end of this year.

The domestic regulation is generally considered to be beneficial to the development of the domestic generative AI industry, and the securities market reflects the expectations of the outside world.

After several weeks of decline, AI concept stocks rose again on July 14. 360, HKUST Xunfei, and Kunlun Wanwei rose by 2.46%, 2.71%, and 4.04% respectively. Baidu rose 2.35% in the US stock market. %.

According to the CITIC Securities Research Report, the development of artificial intelligence application layers is expected to be promoted, such as games, tool-type and entertainment-type applications. At the same time, the policy also emphasizes the promotion of generative artificial intelligence infrastructure and the construction of public training data resource platforms, which are expected to drive the development of companies related to data and models.

Open the "volume" application in the post-big model era

In addition to the promulgation of the "Measures" at the management level, many well-known technology companies are also participating in the formulation of the standardization of the development of the artificial intelligence industry.

At the just-concluded 2023 World Artificial Intelligence Conference, the National Artificial Intelligence Standardization General Group under the guidance of the National Standards Committee announced that the Shanghai Artificial Intelligence Laboratory, Baidu, 360, Huawei, Ali and other companies will jointly serve as my country's first large-scale model standardization task force Team leader, and officially launched the formulation of national standards for large-scale model testing.

According to the requirements, the above-mentioned units need to give full play to their theoretical and practical experience in the field of large-scale models, and provide standardized research topic selection suggestions around outstanding issues in the fields of large-scale model software and hardware bases, key technologies, and industrial applications.

Regulatory norms and industry standards were established almost at the same time, promoting the transition of Chinese generative AI companies to the "regular army", and potential competition and reshuffle will also unfold together.

Since ChatGPT triggered a wave of large-scale models, the arms race of Chinese companies in large-scale models has been particularly fierce. According to the "China Artificial Intelligence Large-scale Model Map Research Report" released by Zhao Zhiyun, director of the China Institute of Science and Technology Information in May this year, according to incomplete statistics, China has released 79 large-scale models with parameters above 1 billion.

The "Hundred Models War" is in full swing, but the excitement caused by the large models is also waning day by day. The outside world expects that under the background that regulatory authorities encourage generative AI to enter the application side, AI companies can come up with something different.

Although there is still a lot of room for optimization of large models, it is no exaggeration to say that the post-large model era has arrived ahead of schedule. The current market has long been tired of the large models produced by repeated wheel creation, which means that the transformation of large models to existing business scenarios will become a new outlet.

Which tracks are expected to be the first to be supported by generative AI? Kaiyuan Securities believes that games, e-commerce and education will bear the brunt.

From production to subsequent player experience, the game industry has a trend of being revolutionized by AIGC. Whether it is game plot design or character original painting design, generative AI can significantly improve efficiency, and the NetEase game "Naishuihan" also introduces AI into the game, allowing NPCs to naturally talk to players and output different interactive experiences. In the long run, AIGC will continue to improve the content richness of the game.

The e-commerce track is also a stage where AI can flex its muscles. Kaiyuan Securities stated that the “AI shopping guide” launched by the leading shopping guide platform with UGC content and data advantages based on large models may promote traffic and revenue growth, and “AI+digital people” is expected to open up a broader space for commercialization.

In the field of education, AIGC technology will accelerate the growth of the youth AI training market. The combination of massive teaching content and large models can make AI a teacher of a specific subject, and the market for AI-assisted teaching is huge. From a specific example, the driving school Dongfang Fashion has enabled AI coaches to guide students to practice driving in real time in the form of voice in the car.

In addition, industries such as film and television and media that focus on content production will continue to be transformed by AIGC. It is expected that in the future, a large number of AI productivity tools based on large models and oriented to vertical industries will be released to assist humans in efficiently completing content creation.

From volume models to volume applications, the wave of generative AI will also flow from giant companies to creative small and medium-sized companies. When large models are combined with creativity, new applications that are completely different from dialogue robots may be born, which will be more unique. of business stories are brewing.