China’s AIGC industry in the future

AIGC (AI-Generated Content) is essentially a form of content production, that is, AI automatically generates content. It is based on deep learning technology and inputs data for the AI to find patterns and generalize appropriately in order to generate content. At present, AI has already achieved the ability to generate text, audio, images, and videos. The basic generation-based AI primarily features textual modalities, with high market interest in audio, image, and video modalities. Text-to-image generation utilizes CLIP as a main training neural network model, where both texts and images are broken down into encoders for mapping, completing the training process. Similarly, text-to-audio also follows similar training patterns. As large models upgrade multi-modal abilities, text-to-video rapidly develops. Following the integration of text-to-image capabilities into various large models, text-to-video has become a new trend in application for multi-modal large models. Recently, several manufacturers have released products or updated their offerings to significantly improve text-to-video effects.

The model layer has two main features, high research barriers and high operating costs. On the one hand, from the perspective of data foundations and training costs, it takes significant amounts of data and ample computing power to develop models, on the other hand, from the perspective of running costs, models require strong computing support. At present, the operation cost of the application layer is relatively low, and it is suitable for enterprises with different expectations in terms of underlying algorithm capabilities. Large companies have advantages in both data and funding, making them well-equipped to handle the model layer. Currently, enterprises are enjoying lower API calls costs for models from the industry as a whole, which helps keep operation costs low, making this an ideal situation for startups.

As the maturity of large-model technology increases, larger models will provide powerful capabilities for AI Agents. The combination of Agent and a large model is expected to build intelligent entities with autonomous thinking, decision-making, and execution capabilities that can further enhance the application capabilities of large models. In the field of artificial intelligence, an AI agent is considered an artificially created entity capable of using sensors to perceive its surroundings, making decisions, and performing actions through actuators. Compared to human-AI interaction patterns, AI agents are more independent than current widely used Copilot modes, which can autonomously call upon resources to complete tasks while humans act as supervisors and evaluators. With a wider range of application, AI agents can handle multiple tasks in different fields and perform various functions; with more natural and flexible communication methods, they can understand complex natural language instructions, engage in intelligent dialogues with users, and provide better services.

From a global perspective, the AI industry experienced a slight decline in growth rate last year after a strong rebound in capital market funding. The generative AI model attracted around $14 billion in investment from the first half of this year alone, demonstrating its popularity and strong demand in the market. However, most companies are still at an early stage in fundraising rounds, with significant future financing needs remaining.

Looking at China’s AIGC industry, similar trends can be observed. The peak was reached in 2021, but the impact of the pandemic led to a decline in funding amounts last year. Since then, there has been a steady rebound in this sector. Given that AI-generated content itself demands considerable investment, it is expected that AIGC will continue to attract attention from investors and maintain its position as a promising area in the capital market through 2024.

The Chinese AIGC industry has already developed into two main types of business models. The first type is focused on serving C-end users, providing a variety of content forms mainly categorized by modalities such as text generation, image generation, audio generation, video generation, and virtual world creation. Meanwhile, the second type targets B2B enterprise clients with more specialized services in specific fields, currently focusing on areas like games, media/film, e-commerce, and advertising marketing that place high demands on content production. In the future, it is expected that AIGC will continue to expand its industry chain and deepen its impact on various industries by expanding commercial scenarios.

The gaming industry possesses features such as high interactivity and content creativity, making it one of the most likely sectors to benefit from the impact of AIGC. In the game development field, the application of AIGC is becoming increasingly widespread, with its potential benefits including reduced costs for developers, improved efficiency, and innovation in gameplay. Furthermore, players can expect more rich and realistic gaming experiences as a result of adopting AIGC technology. Currently, an increasing number of domestic game companies are incorporating AIGC techniques into their workflow processes.

In the media and entertainment sector, AIGC has been applied throughout the early-stage planning, mid-stage production, and late-stage promotion stages. Currently, AIGC is already quite mature in its application during the mid-stage production process, allowing for virtual scene generation, automatic AI editing, AI face replacement, intelligent annotation of image quality, and providing a more clear and efficient material management method for production teams.

In healthcare diagnosis, AIGC has reached an assisted diagnostic stage where it can participate in disease screening, medical imaging analysis, and therapeutic work to improve service efficiency. In the pharmaceutical field, AIGC is mainly applied during the drug development process that involves activities such as target discovery, compound design, clinical research, etc., enabling AIGC to perform tasks like target finding, compound synthesis selection, and clinical trial participant screening.

AIGC has been applied in the financial industry to cover various aspects of front-end, middle-end, and back-end marketing, research projects, product design, risk control compliance, customer service, operation management, etc. It is supported by five major AI technologies at its foundation.

AIGC can enable professional creators and individuals to freely express their creativity, reduce the threshold for content production, and bring a large amount of content supply. For the still-experimental metaverse world, AIGC technology also brings about possibilities in solving problems related to creating content for the metaverse, playing a key role in constructing the foundation of the metaverse.

As an important branch within the field of artificial intelligence, AIGC has promising future prospects. Technological innovation and expansion of application scenarios will be the main driving forces. Cross-sectoral cooperation and improvement of legal regulations will provide guarantees for industry development.