During a panel discussion at the Boao Forum for Asia in Hainan province, Liu Cong, the vice president of Chinese AI firm iFlytek, acknowledged the existing gap between Chinese and US leaders in generative AI. He stressed the importance of focusing efforts on developing independently owned and controlled hardware and software, particularly in large language models (LLMs).
LLMs are deep-learning AI algorithms powering chatbots like ChatGPT, enabling tasks such as recognition, summarization, translation, prediction, and content generation using extensive datasets.
While Chinese firms like iFlytek have launched their own chatbots, there’s recognition of the gap between Chinese and US products. Export restrictions on semiconductor chips, crucial for processing LLMs, pose a significant hurdle to Chinese AI development. The US ban on selling powerful chips to China, citing national security concerns, has prompted a need for Chinese firms to focus on hardware and software development.
Zeng Yi, a professor at the Chinese Academy of Sciences and head of the China Electronics Corporation, emphasized the necessity for Chinese firms to develop new hardware and software, anticipating potential decoupling from US technologies.
Despite China’s perceived lead in areas like facial recognition and autonomous driving, the introduction of text-to-video generator Sora by OpenAI has widened the gap, underscoring the importance of computing power.
Yuan Hui, founder and CEO of AI Chatbot firm Xiao-I, stressed the need for China to strengthen its core technology development, moving beyond applications to focus on foundational technology. Christopher Thomas of the Brookings Institution echoed similar sentiments, highlighting the dominance of the US in computing and urging against duplication of efforts.
Yuan expressed optimism for breakthroughs, emphasizing the inevitability of competition and the need for continuous innovation. However, challenges such as the US chip ban and domestic constraints like computing costs and censorship remain.
Zeng identified risk aversion as a significant challenge to China’s AI development, questioning the willingness to challenge prevailing norms and invest in exploration. Overcoming this challenge, he suggested, is crucial for the advancement of artificial intelligence in China.