SCMP: China’s ChatGPT quandary the result of a lack of vision and fundamental research, say researchers

[Hamartia Antidote]

Researcher blames China’s lack of a ChatGPT alternative on lack of vision

A lack of visionary leaders and funding contributed to China falling behind in large language models, according to a Chinese Academy of Sciences researcher.

www.scmp.com

• A lack of visionary leaders and funding contributed to China falling behind in large language models, according to a Chinese Academy of Sciences researcher
• The question of why a Chinese company has not produced something like ChatGPT has plagued the domestic AI industry since the hit chatbot was released last year

The overnight success of ChatGPT has left many in China wondering why the domestic artificial intelligence industry did not beat OpenAI to the punch. Photo: Shutterstock

A researcher at one of China’s top scientific institutes has suggested a lack of vision may have contributed to China falling behind in ChatGPT-like artificial intelligence (AI). The verdict comes as questions abound about why the country appeared to be caught off guard by developments in large language models that turned the chatbot from San Francisco-based start-up OpenAI into a global sensation and ignited intense competition around the technology.

“The reasons are aplenty, from the need for visionary leaders, top-notch teams and sizeable funding,” Bao Yungang, a researcher at the Institute of Computing Technology at the Chinese Academy of Sciences (CAS), wrote on microblogging platform Weibo. He also highlighted OpenAI’s shift in 2019 from a non-profit company to one with profit capped at 100 times investments as a major contributor to its success.

The overnight success of ChatGPT has spurred soul-searching in China, which has been a leader in some areas of AI but fallen behind in language models capable of answering complex queries and producing myriad types of content. Bao said that similar questions about falling behind could be asked of leading research schools in the US like Stanford University and the Massachusetts Institute of Technology.

Chinese Academy of Sciences researcher Bao Yungang. Photo: Handout

China has other challenges, though. One is an unwillingness to invest in fundamental research, according to Liang Zheng, deputy head at the Tsinghua University’s China tech policy research centre.


“So far there is literally no Chinese company investing heavily in fundamental research. No one is doing the step from 0 to 1,” Liang told the local newspaper Economic Observer last month. China needs to “take some time to cultivate a new market environment supportive of innovation, encouraging those with a long-term mindset to work on fundamental and explorative work, which could eventually lead to something meaningful”, he added.

Since the launch of ChatGPT in November, many Big Tech firms have rushed to put out or show progress in developing their own similar products. In Silicon Valley, Google rushed out a preview of its AI-powered chatbot Bard, receiving a cool response.

A number of tech giants in China have come out publicly as working on ChatGPT-like services. The biggest contender is Baidu, which released its Ernie Bot last month.

Bao attributed the inability of other companies to produce a true ChatGPT rival to Microsoft-backed OpenAI as “finding a lot of small wins and multiplying them together”. People in China should have more confidence in domestic companies to be able to rival OpenAI over time, Bao said.

“We can’t expect [Ernie Bot] to catch up to GPT-4 immediately,” Bao wrote in his post, referring to the latest GPT language model that OpenAI released last month. “But as long as Baidu keeps iterating the service to solve hundreds and even thousands of problems, believe it will become better.”

While generally considered to lag behind the US in AI, China is a leading producer of related research. The country produces more AI journal, conference and repository publications than anywhere else, according to the latest Artificial Intelligence Index Report released by Stanford University this week. In total AI funding, China was found to be second only to the US, although it invested a third as much at US$13.4 billion compared with US$47.4 billion in the US.

However, the US supplies more than half of the world’s most significant machine learning systems, with 285 coming from the country compared with 49 from China, according to the report. Experts have pointed to challenges from censorship in China as possible factors contributing to issues in training models like those powering ChatGPT, which can give unpredictable and often inaccurate responses to questions.


Ben Jiang

 

[letsrock]

ChatGPT is being way overhyped for now. But it is still a technology that must be invested in and will pay crucial role in future.

To make full use of chatgpt the society must also generate huge content in every area - sciences, government policy, research reports, business processes. If china is not already disseminating content in digital format in huge volumes and content includes business processes etc then chatgpt is just entertainment.

 

[leviathan]

Can ChatGPT save your SVB? How come the SVB CEO did ask ChatGPT for advice or tip?

 

[Mista]

Well, their tech companies were grappling with Xi's 'common prosperity' tech crackdown for the past few years. And they don't have access to the most advanced chips to train their AI, which is a competitive disadvantage.

Nvidia tweaks flagship H100 chip for export to China as H800

Nvidia Corp, the US semiconductor designer that dominates the market for artificial intelligence (AI) chips, said it has modified its flagship product into a version that is legal to export to China.

www.scmp.com