IT sector leaders in Hong Kong say they’re excited by the possibilities of emerging artificial intelligence technology such as OpenAI’s ChatGPT, but one entrepreneur doesn’t expect cautious local businesses to make use of it any time soon.

In the two months since its release to the public, the chatbot has become a global sensation for its ability to generate comments and answers in a vast range of styles on just about any topic. While it isn’t officially available in the SAR yet, people in the sector have been giving it a try using virtual private networks, or VPNs.

Duncan Chiu, who represents the IT sector in Legco, says only a handful of companies worldwide have the resources to build a technology like ChatGPT from scratch, but there’s nothing to stop Hong Kong businesses developing AI solutions using OpenAI’s technology or that of its rivals.

"Companies like Google, or maybe in China – Baidu, Tencent [and] Alibaba; those are the very few companies [that] have the network and money to build [a conversational AI model]," Chiu said.

"But below that... there could be many applications where people can think about applying [the technology] to work on particular industries."

He suggested the chatbot could be put to use in tasks such as copywriting and even elements of journalism, and the fact it isn’t officially available here yet should not stop businesses here from making use of it.

He said people could access the chatbot using a VPN and work out tools and applications of their own that could then be sold as customised products in Hong Kong.

One businessman who is already making the most of ChatGPT is Kinni Mew, the founder and lead engineer of Mindlayer, a local company using AI chats for customer service functions.

Mew says the technology itself is not anything new, but ChatGPT performs far better than any previous chatbot. That is generating excitement in Hong Kong’s AI community.

"This is the first time that this kind of generative model can produce such high quality," he said.

"Google and another big tech also released similar tools like this. But the performance is not as good," Mew added.

Google’s Bard chatbot has been in the spotlight for all the wrong reasons. The release last Monday of a promotional video in which Bard gave inaccurate information in response to a question has been blamed for wiping US$100 billion off the value of the tech giant’s parent, Alphabet.

And the Bard blunder underlines a point made by both Chiu and Mew about the limitations of the technology.

"It does not have any computation or logical thinking. It's based on the past data [that] have been fed into the model," Mew said.

"If somebody is trying to train a model with a lot of misleading information, the model itself gives you [the wrong] answer instead of the correct [one]."

And Mew said clients in Hong Kong were more likely to be concerned about the possible negative consequences of something going wrong with AI technology than those in other territories such as the United States, where entrepreneurs were more likely to focus on the upside.

"They're more open to this kind of tool. They can take that risk. They can use that kind of generic model to let the model itself generate the content without considering too much about the restrictions," he said.

By contrast, Hong Kong companies "would rather give up those good things to prevent those bad things happening".

"For example, if I[a user] asked how much the application fee is, the correct answer may be 20. But the chatbot [may] give you a random answer like 100. This kind of wrong result will really discourage the company to try new things," he said.

"Innovation should take [the] risk," he said adding that the AI can be inaccurate but: "Such a situation may be just only one percent of the case. But 99 percent of the time, you reduce the cost, you save more time, save more effort on solving the problem."

Chiu, the lawmaker, said it was important for Hong Kong to train and educate people on the development of AI, saying: "There's no going back on technology. It will only get better and better."

Mew said he and his colleagues were using the chatbot to code, extract key concepts from lengthy articles, and even summarise meetings with clients.

"[My colleague] used the tools to generate the source code template as a starting template. Then, my colleague will work on it, and try to modify the code. That can speed up the programming effort as well," he said.

And Chiu stressed that using AI tools was about enhancing how work is done, for example by saving time, rather than removing the human element, adding: "AI will not replace humans, but AI will replace humans that do not use AI.”