Chinese Are Undergoing a New Evolution With AI

cs_opinion_img
If a country wants to become an AI power, hardware, software and talent are important, but China’s unique competitive advantage comes from an underlying logic with thousands of years of history.
May 28, 2025
author_image
Editor-in-Chief for China Currents and Top Picks; Wave Media Correspondent
Click Register
Register
Try Premium Member
for Free with a 7-Day Trial
Click Register
Register
Try Premium Member for Free with a 7-Day Trial

Your Chinese friends might have been keeping a secret from you – every Chinese person is secretly qualified to be a designer, 3D modeller, or even a film director. Don’t believe me? Let me show you right now.

Imagine you’ve just come up with a brilliant idea for a game character, but you’ve never learned to draw or model. What should you do?

Just describe your idea to Hunyuan 3D in plain text, and it will automatically generate a 3D model of your character. Refine the design until it matches your vision perfectly, then download the model file. You can 3D print it, or import it into professional modelling software for further development.

See? No drawing skills are required at all. These steps are all completed through an open-source AI developed by Chinese technology company Tencent, it’s free in China. That’s a chance to experience the thrill of being a designer for everyday people, and also a serious productivity booster for professionals.

If you’re a gamer, you’ve probably noticed this: some games can handle massive battles with hundreds of characters smoothly, while others lag even with only a few characters.

That’s because what strains your device isn’t the number of characters but the number of surfaces being rendered. A cube the size of Earth still only has six sides, while a simple basketball may require calculating 200 to 500 faces. To ensure smooth performance on lower-end systems, developers typically start with a high-resolution model, and then create medium- and low-poly versions. For example, the developers of Black Myth: Wukong started with a carving model with 1.2 billion polygons and reduced it to 20 million before using it in-game. This process—called Level of Detail (LOD)—is time- and resource-intensive. However, according to Tencent’s AI engineer, Hunyuan 3D-2 can automate the LOD process and even rig skeletal animations. For 3D animation studios and game developers, this is game-changing.

Traditionally, creating a single game character costs around 12,000 yuan and takes half a day. With Hunyuan, the same task takes just one minute and costs about 0.5 yuan—cutting modelling costs by an astonishing 99.996%. In fact, this technology has far more promising use cases beyond gaming.

The first is video generation

While models like Sora and Dream Avatar are already highly advanced, they still produce 2D videos. When it comes to complex camera movements or multi-angle shots of the same scene, these models are prone to generating visual inconsistencies. Repeated regenerations during the debugging process also consume massive amounts of tokens.

In contrast, Hunyuan can generate 3D scenes from just a photo—or even a brief description. The advantage of generating video from 3D models is that once a model is created, it can be reused from any angle while maintaining consistency. It allows for precise control over character poses, camera positioning, and lighting—much like stop-motion animation. This could revolutionize the film and advertising industries and empower independent creators to bring their boldest visions to life.

The second major application lies in robotics and autonomous vehicle training

In February, Unitree’s robot dogs captured widespread attention as they gracefully navigated gravel, bushes, and snow, even climbing upstream waterfalls—demonstrating world-class gait algorithms. The rapid progress in robotics owes much to virtual training environments that have dramatically accelerated learning.

Previously, robot dogs might get stuck on common obstacles like tables—not due to poor sensors, but due to insufficient experience with various table designs. Engineers can’t possibly collect every type of table for training. But in virtual environments, Hunyuan can randomly generate tens of thousands of different tables, allowing multiple robots to train simultaneously in parkour-style environments. This significantly boosts AI robustness. The same approach can be used to train autopilot systems—enhancing their ability to handle rain, snow, and construction zones, improving safety.

From gaming to autopilot, Hunyuan seems to be applicable to every lucrative sector. And yet, Tencent has chosen to make it open source and free to the public. They’ve even generously shared their computing power and user base with other tech companies through the Tencent Yuanbao app, enabling more Chinese users to access DeepSeek models. Why? Is Tencent a charitable organization in disguise?

The simplest explanation might be this: it is a Chinese company.

On one hand, open-source development is a major trend in China’s AI industry. Developers have widely recognized that advancing better algorithms requires more feedback, and feedback comes from more users. High paywalls may offer short-term profits, but they also limit the growth of the number of users and stifle long-term innovation. Through the successes of DeepSeek, Hunyuan, and Qwen, China has continuously pushed the boundaries of open-source AI, fostering a culture of positive-sum competition. Even OpenAI announced in April 2025 that it would reintroduce open-source models, validating China’s strategic direction.

Chatbot Arena LLM Leaderboard,May 22, 2025

On the other hand, Tencent has long upheld a mission of “Tech for Good.” Back in 2007, it established the Tencent Charity Foundation—the first public welfare foundation initiated by an internet company in China. Over the years, Tencent has applied technology to causes ranging from wildlife protection and carbon reduction to elder care. Teaching more people to embrace the joy of creation is just one small part of that broader mission.

As early as 770 BCE, Chinese philosopher Guan Zhong advocated for a “People-first” approach. In 2023, President Xi Jinping reaffirmed this idea in the Global AI Governance Initiative, emphasizing that AI must be “people-oriented and beneficial to humanity.” As a Chinese company, Tencent naturally carries this legacy forward.

Therefore, the reason why Chinese people are among the first to benefit from this new wave of open-source AI evolution is not out of a desire to dominate others—but because an ancient civilization, as it strides into the future, insists on leaving no one behind.

Editor: Charriot Zhai

VIEWS BY

author_image
Editor-in-Chief for China Currents and Top Picks; Wave Media Correspondent
Share This Post

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Comment
Cancel

  1. G

    Some Guan Zhong aphorisms:
    “When the granaries are full, the people understand propriety; when food and clothing are sufficient, they know honor and shame.”
    “A ruler who gains the hearts of the people prospers; one who loses them perishes.”
    “Governance begins by enriching the people.”

    Show more
    Show less
    likednot_liked 0likednot_liked 0Reply