 "Despite their aspirations to lead in generative AI, Chinese firms heavily rely on underlying systems from the United States, mainly open-source models like Meta's LLaMA")
Chinese tech companies would like people to believe that their AI bots and applications are home-grown and completely indigenously developed. However, a new report by The New York Times seems to indicate that China’s tech sector is still largely dependent on the US and its technology for their “home-grown” AI models.
As restrictive as China’s AI landscape is, several Chinese tech companie have managed to develop their own AI bots and applications. What has been surprising is the way several startups in China have gone toe to toe against major, established tech companies, some of which have the backing of the Chinese Communist Party. Several fledgling Chinese startups like 01.AI, had a blast thanks to open-source artificial intelligence systems, just a year after the release of ChatGPT.
Founded by renowned investor and technologist Kai-Fu Lee, the company, despite being only eight months old, boasted a staggering $1 billion valuation, backed by deep-pocketed investors.
However, like many other AI startups in China, 01.AI’s rise to prominence came with a twist - it leveraged technology from Meta’s LLaMA, presenting its AI system as an alternative to existing options.
Despite their aspirations to lead in generative AI, Chinese firms heavily rely on underlying systems from the United States, mainly open-source models like Meta’s LLaMA.
The fact that most startups are relying on solutions from American companies goes on to show that tech companies in China are lagging significantly behind the US in generative AI. This is deeply concerning among industry insiders about the trajectory of technological competition between the two nations.
Impact Shorts
View AllChina has placed significant emphasis on advancing AI technology, particularly highlighted by the victories of AlphaGo over leading Go players in 2016 and 2017.
This success prompted Chinese policymakers to establish an ambitious agenda aimed at global technological leadership by 2030. To achieve this, substantial investments were promised to support AI research and development within both academic and corporate sectors.
However, in November 2022, when OpenAI unveiled ChatGPT, numerous Chinese enterprises found themselves constrained by stringent regulatory measures imposed by the Beijing government.
Additionally, Chinese tech giants also have to contend with new and arbitrary regulations governing the training of AI models. These regulations imposed constraints on the datasets permissible for training AI models, delineated acceptable applications, and mandated the registration of AI models with the government.
These regulations discouraged innovation without official approval, hindering the ability of tech firms to explore new ideas freely. Moreover, censorship guidelines aimed at controlling public discourse and minimizing dissent against the Chinese Communist Party added further challenges for technology companies in China.
“When Chinese companies are leveraging American open-source technologies to play catch-up, the questions become very complicated — wrapped up in issues of national security and geopolitics,” said Oren Etzioni, a University of Washington professor and the founder of TrueMedia.org, a nonprofit working to identify disinformation online in political campaigns.
Some experts believe that China could be at least two to three years behind the US in generative AI.
The implications of this AI race are substantial, potentially tipping the global technological balance of power. While Chinese firms scramble to catch up by adopting U.S. open-source AI models, Washington finds itself in a complex position, balancing strategic interests with technological sharing.
This reliance on U.S. AI systems has sparked deeper questions about China’s innovation model, particularly as it navigates authoritarian controls. Oren Etzioni, a University of Washington professor, emphasized the complexities intertwined with issues of national security and geopolitics.
According to Yiran Chen, a John Cocke Distinguished Professor of Electrical and Computer Engineering at Duke University, tech investors in China have been prioritizing quick returns from AI investments. This focus has led to capital flowing towards easily implementable applications rather than more ambitious endeavors centered on fundamental research. Chen notes that as much as 50% of China’s AI investment has been directed towards computer vision technology, primarily utilized for surveillance purposes, rather than towards the development of foundational models for generative AI.
Recently, major players like Baidu, Alibaba, along with unexpected entrants such as the dairy company Mengniu and the tutoring firm TAL Education, have all joined the race in developing generative AI technology in China. This surge of activity has led Chinese media to dub the phenomenon as “the battle of 100 models,” signifying the intense competition and proliferation of initiatives in the field.
Yet, challenges persist, including US restrictions on AI chip sales to China. Despite stockpiling chips, Chinese companies face hurdles in sustaining operations.
While China grapples with these obstacles, the unveiling of OpenAI’s Sora, a tool for generating Hollywood-like videos, highlights the ongoing dynamism in the global AI landscape, where innovation remains fiercely competitive.
(With inputs from agencies)