Tencent launches Hunyuan corporate AI model amid Chinese competition

Chinese tech giant Tencent is launching its “Hunyuan” AI model for commercial use at an annual summit on Thursday, Dawson Tong, CEO of Tencent’s Cloud and Intelligent Industries Group, told CNBC in an exclusive interview before the event.

The news comes days after Baidu revealed a slew of AI-powered apps on Tuesday in the wake of more supportive regulation.

Tencent said it was internally testing its Hunyuan AI model on advertising and financial technology. The company said in an online post that the gaming and social media giant is also set to launch an AI-powered chatbot on Thursday.

Tong told CNBC that Tencent is working to integrate Hunyuan’s capabilities with its existing video conferencing and social media products.

The company operates WeChat, a messaging and payments app widely used in China, and videoconferencing platform Tencent Meeting.

Baidu and several other Chinese companies have been given the go-ahead in the past few weeks to release AI chatbots to the public.

Similar to ChatGPT, the bots claim to respond to queries in a conversational, human-like manner, but primarily in Chinese. Some, like Baidu’s Ernie bot, also convert text into images and video with the help of plug-ins.

OpenAI’s ChatGPT is not officially available in China. The chatbot versions follow new Chinese regulations on generative AI that took effect on August 15th.

When asked about the rules, Tung noted that such artificial intelligence is so new that no one knows how much it will affect society.

“It would be wise to put some guardrails in place,” he said. This, he said, will help ensure that the technology or services provided are of high enough quality so that they do not create and distribute false information.

Chinese authorities said the “temporary” rules that took effect last month would not apply to companies developing AI technology as long as the product was not available to the general public.

This is more relaxed than the draft in April which said the upcoming rules would apply even in the research phase.

And while Beijing has shown itself to be more supportive of generative AI than initially feared, Chinese companies also face US restrictions on access to advanced semiconductors. The latest versions of high-tech chips, known as graphics processing units (GPUs), allow companies to train AI models.

“The restrictions we face will hinder progress and speed of development,” Tung told CNBC when asked about the US restrictions.

He noted that the demand for computing power in general far exceeds the supply in China. To mitigate the shortage, he said, the companies are “focusing on specific use cases, building models of the right size.”

“We hope that the supply of GPU computing will be greater in the coming months, so the development of these technologies can become faster.”

Tencent is just one of many companies in China — from start-ups to phone maker Huawei — that have rushed to announce AI products this year. In August, Alibaba announced that it would open its AI model to third-party developers.

Tencent’s Tung said AI requires industry-specific training so the technology can generate value. He listed commercial use cases in the areas of tourism, finance, public services and customer service.

“We believe that many different customers will, in fact, benefit most by leveraging open source models and using their enterprise data to train their own models for very specific needs in their industrial use cases,” he said.

He said this customized use can also help protect data.