Collaboration key to success

Ever since the launch of OpenAI’s ChatGPT late last year, Western and Chinese tech titans have been scrambling to roll out their own artificial intelligence chatbots, with businesses eager to cash in on the AI revolution.

However, while larger firms can easily use ChatGPT in a myriad ways to benefit their businesses, smaller enterprises will find it difficult to take advantage of these innovations as they lack technical prowess to do so, an expert says.

Therefore, they would be better off joining hands with AI specialists, says Derek Ng Tak-chuen, the director of Hong Kong startup Parami, which specializes in AI-based chatbots and solutions.

Ng believes that some firms are hesitant to integrate ChatGPT into their business model due to security risks or the possibility that the chatbots could give them wrong answers or solutions.

When a company starts using ChatGPT, it is in effect transferring part of its business process to the AI tool to complete, but sometimes this tools may not be reliable enough in specific scenarios, according to Ng.

He points out that it is not hard for internet giants to create ChatGPT-like products as their business models give them access to massive amounts of data.

But he adds that data quality is as crucial as data volume in making a generated AI model.

Earlier this year, Chinese search engine giant Baidu (9888) unveiled its ChatGPT-like product called Ernie Bot but its launch disappointed investors, with chief executive Robin Li Yanhong admitting that the chatbot was “not perfect.”

“We have analyzed Baidu’s language model before, and its quality is not very good,” says Ng.

“The reason why ChatGPT is relatively good is that there are many people working to maintain the data quality,” he explains.

But ChatGPT has also come in for flak, with some netizens revealing that the bot was unable to give correct answers when given math questions in a roundabout way. And Google’s parent Alphabet lost US$100 billion (HK$780 billion) in market value in one day in February after its new chatbot Bard shared inaccurate information in a promotional video.

Nevertheless, businesses and developers are still pushing ahead.

Chinese technology giant Alibaba (9988) plans to roll out its own ChatGPT-style product called Tongyi, as does Tencent (0700), while telecom giants have joined the race as well with China Telecom (0728) developing its own ChatGPT-like model and China Unicom (0762) also stating its intentions on AI.

Ng says if telcos cooperate with tech titans, they will be more likely to succeed.

For example, if China Mobile combines its sound data with Baidu’s text data, it may be able to create a multimedia GPT, he points out.

However, China’s internet giants may also be under pressure from government regulators when developing AI in the future as they have access to vast amounts of data, Ng adds.

He believes that a vertical application scenario between ChatGPT and businesses is essential in commercializing such a tool.

As a general-purpose AI system, ChatGPT can be used in many scenarios, such as daily work routine, customer service, content creation, and writing codes, significantly improving work efficiency.

OpenAI has now released application programming interface or API access to ChatGPT in which companies can use API to create a custom model based on their business logic.

But the ChatGPT business model is still being explored as firms may not find a way to integrate it into their operations, Ng says. Also, he points out that any product needs a measurable indicator which people can use to rate it, but this still does not exist for AI tools, and this could be the biggest challenge for the commercialization of tools like ChatGPT.