ChatGPT could spur a gold rush for Chinese firms

The furore created by San Francisco start-up OpenAI’s ChatGPT bot has set off another round in the artificial intelligence (AI) arms race among global tech giants.

And while ChatGPT originated in the US, Chinese boutique investment bank Lighthouse Capital is betting that the ability of Chinese companies to turn innovative concepts into usable products quickly, could give them a leg up.

“While [many] innovations come from overseas markets in the first place, Chinese companies have the capability to turn innovations into excellent products,” Xie Xuzhang, executive director at Lighthouse Capital, said in an interview with the Post on Wednesday.

“In the long run we believe a lot of Chinese companies with tech and product strength will emerge to become industry leaders [in this space],” said Xie, adding that local companies would certainly dominate the domestic market due to their access to Chinese text materials and compliance knowledge.

ChatGPT, which stands for Chat Generative Pre-Trained Transformer, is a conversational bot that Microsoft-backed OpenAI unveiled in November. It is built on top of OpenAI’s GPT-3 family of large language models and has been fine-tuned using both supervised and reinforcement learning techniques.

The bot can understand sophisticated prompts and generates human-like responses. It has continued to learn and improve its responses since coming on line, interacting with people and receiving more information. The enthusiastic public response to ChatGPT has spurred Google and Microsoft to bake similar technology into their new product offerings.

There has been a like-minded reaction in China, with tech firms such as Baidu – which has put years of investment into AI and its applications – raising their bets on local competitors to ChatGPT, even though existing products are still based on older GPT-2 models.

Lighthouse’s Xie accepts that overseas companies will likely still reign on the global AI stage when it comes to cutting-edge technology.
Building up local competitors to ChatGPT will face challenges, such as gaps in training models, the availability of data sets, as well as the uniqueness of the Chinese language itself.

Xie’s comments echo a view recently expressed by Ding Wenxuan, a professor of AI and business analytics at France’s Emlyon Business School.

Ding said in a recent online webinar hosted by local media outlet iFeng that Chinese is a pictographic language with polysemy – meaning that words or phrases can have multiple meanings depending on the context in which they are expressed. In contrast, Ding said the English language is descriptive, making it easier for language models to understand.

However, while ChatGPT has stirred scientific debate within the tech community, its commercial value remains unproven to date and human intervention is still required. While the bot is chatty and knowledgeable, and is able to summarise a report or draft an email, it remains an entertainment application at this point, said Xie.

OpenAI has said ChatGPT – at least for now – should not be used in areas where tolerance for inaccuracies is low, and that it is better suited for tasks that involve calling up facts rather than those that require “logical reasoning”. For example, it can tabulate market data for finance professionals but cannot recommend an ideal investment portfolio.

Xie said that when making investment decisions on local ChatGPT candidates, judgments will centre on the team’s technology strength and – more importantly – how that fits into a clear business scenario, as well as ease of access to data in relevant areas.

Xie sees three clusters of investment opportunities arising from the ChatGPT gold rush: underlying language models; the middleware that connects the models and upstream applications; and opportunities at the application level – which will be a key area for Chinese start-ups and firms. South China Morning Post