According to The New York Times, the Chinese company 01.AI has developed the Yi-Large Global SOTA LLM, which is based on open-source technology and has achieved performance levels comparable to American models like GPT-4o and Llama-3. Additionally, a video generation tool named Kling is also available in China, functioning similarly to OpenAI’s Sora, but with broader access for users.

The AI revolution began in the U.S. with the launch of ChatGPT in late 2022. However, Chinese firms are rapidly advancing, leveraging open-source AI code that enables users to enhance and innovate upon existing technologies. Open-source frameworks are foundational to many technological advancements, from operating systems like Linux to new architectures like RISC-V.

While many American companies lead in AI innovation and have shared their data, concerns regarding the ethical implications of AI have led to hesitations. In contrast, Chinese companies have been more liberal in making their technologies accessible to the public and sharing code with developers and institutions.

This collaborative approach has significantly accelerated AI development in China, allowing companies that were once criticized for using open-source code to now excel in AI benchmarks.

Despite the stringent restrictions imposed by the U.S. on China’s technological growth, such as bans on high-end chips, Chinese firms are making strides in software development. Experts suggest that while hardware parity with the U.S. may take a decade, Chinese tech companies are catching up in software capabilities.

However, the U.S. government is considering legislation to tighten control over the export of American AI technologies, which could hinder the open-source development that has propelled U.S. AI leadership. Clément Delangue, CEO of Hugging Face, remarked, “Open-source AI is the foundation of AI development,” highlighting the collaborative efforts that have positioned the U.S. as a leader in AI, and suggesting that China may replicate this success.

Moreover, Stanford University’s recent Llama 3-V LLM utilizes open-source code from China, indicating that American researchers are beginning to rely on Chinese AI advancements. Delangue cautions that if the U.S. restricts open-source development, it may lead to a scenario where American developers increasingly depend on Chinese technologies.

Despite the progress made by Chinese firms in open-source AI, Professor Yiran Chen from Duke University believes they still face challenges. “Chinese companies excel at replicating and enhancing existing U.S. technologies,” Chen stated, “but they struggle to create entirely new innovations that could surpass U.S. advancements in the next five to ten years.”

Whether this assessment holds true remains to be seen, but the next decade will be pivotal in determining the trajectory of AI development in both nations.

Jowi Morales is a tech enthusiast with years of experience in the industry, contributing to various tech publications since 2021, focusing on hardware and consumer electronics.