China can use ChatGPT-like tech to upgrade economy but limits remain

China’s application of ChatGPT-style services could face hurdles including content compliance, costs and computing power, some local experts have warned, as the country’s Big Tech firms push ahead with bets on the technology.

“The training of ChatGPT large language models (LLMs) needs significant computing power … investments that easily run into billions of Chinese yuan are beyond what most firms can afford, unless they are gigantic,” said Xiao Xiupeng, a vice-president at Xiao-i, a Nasdaq-listed Chinese AI firm. He was speaking on the sidelines of the Shanghai AI Conference held from April 15-16, according to a report by local media Shanghai Securities News.

As Chinese tech titans like Baidu, Alibaba Group Holding and Tencent Holdings rush to roll out competitor products to ChatGPT, Xiao warned of the challenges in China, including compliance and privacy requirements. He also referred to prohibitive development costs, limits on computing power as well as information accuracy problems with these chatbots.

San Franciso-based OpenAI launched ChatGPT, a conversational bot backed by its GPT LLM last November, to global fanfare as it is able to understand complex questions and respond with humanlike text.

Chinese tech firms have been quick to respond, with Alibaba last week launching Tongyi Qianwen, which means seeking truth from a thousand questions, in Beijing last week, hot on the heels of Ernie Bot from Chinese search and AI giant Baidu. Alibaba owns the South China Morning Post.
Although cautious, Xiao still sees opportunities for Chinese AI efforts in this field, particularly when it comes to leveraging large data troves to provide bespoke solutions for specific industries, such as finance, government services and energy.

Xiao’s comments echo the view of Xie Xuzhang, an investor at Chinese boutique investment bank Lighthouse Capital, who told the Post in February that “while [many] innovations come from overseas markets in the first place, Chinese companies have the capability to turn innovations into excellent products”.

Xie said that from an investor’s perspective, there will be more opportunities for Chinese start-ups and firms at the application level.

Cao Yan, a general manager at Tencent Cloud’s industrial ecosystem arm and a speaker at the Shanghai AI forum, said at the event that the best way for smaller Chinese companies to leverage the prowess of LLMs would be for them to have smaller AI models trickle down from the larger models, and then use them in actual business scenarios.

“The use scenarios will not be done independently by just one company, a symbiotic ecosystem is needed,” he said.

Speakers at the event also noted the importance of AI when it comes to upscaling China’s manufacturing sector, which has been hit by geopolitical tensions and moves by some multinational companies to diversify their supply chains away from China.

Zhang Chengyu, the general secretary at a think tank focusing on the digitisation of companies located in the Yangtze Delta region, gave a speech at the conference on incorporating AI into manufacturing, and how the integration is pivotal to driving the digital transformation at factories when it comes to supply chains, quality control and cost management. South China Morning Post