Tencent releases AI model Hunyuan for businesses amid China competition

Chinese language tech big Tencent is launching its synthetic intelligence mannequin “Hunyuan” for enterprise use at an annual summit on Thursday, Dowson Tong, CEO of the cloud and sensible industries group at Tencent, advised CNBC in an unique interview forward of the occasion.

The information comes days after Baidu revealed a slew of AI-powered purposes on Tuesday within the wake of extra supportive regulation.

Tencent has mentioned it was internally testing its Hunyuan AI mannequin on promoting and fintech. The gaming and social media big can also be set to launch an AI chatbot on Thursday, the corporate mentioned in an internet submit.

Tencent is integrating Hunyuan’s capabilities with its current merchandise for video conferencing and social media, Tong advised CNBC.

The corporate operates WeChat, a extensively used messaging and funds app in China, and video convention platform Tencent Assembly.

Baidu and several other different Chinese language firms acquired the inexperienced gentle in the previous couple of weeks to launch AI-powered chatbots to the general public.

Much like ChatGPT, the bots purport to answer queries in a human-like, conversational style — however primarily in Chinese language. Some, corresponding to Baidu’s Ernie bot, additionally convert textual content to photographs and video, with the assistance of plugins.

OpenAI’s ChatGPT is not formally obtainable in China. The chatbot releases comply with new Chinese language regulation on generative AI that took impact Aug. 15.

When requested concerning the guidelines, Tong identified that such synthetic intelligence is so new that nobody is aware of what impression it’ll have on society.

“It is prudent to place in some guardrails in place,” he mentioned. That may assist be certain that the expertise or the companies being provided are of excessive sufficient high quality so they do not create and distribute false info, he mentioned.

Chinese language authorities mentioned the “interim” guidelines that took impact final month wouldn’t apply to firms growing the AI tech so long as the product was not obtainable to the mass public.

That is extra relaxed than a draft launched in April that mentioned forthcoming guidelines would apply even on the analysis stage.

Whereas Beijing has proven it’s extra supportive of generative AI than initially feared, Chinese language firms additionally face U.S. restrictions on acquiring superior semiconductors. Essentially the most innovative variations of the high-tech chips, referred to as graphics processing models (GPUs), enable firms to coach AI fashions.

“The constraint that we’re going through will hinder the progress, the velocity of growth,” Tong advised CNBC in response to a query about U.S. restrictions.

He famous demand for computing energy general far exceeds provide in China. To mitigate the scarcity, he mentioned firms are “specializing in particular use instances, constructing fashions of the suitable measurement.”

“And we hope that the availability of the GPU compute can be bigger within the coming months, and due to this fact the event of those applied sciences can get sooner.”

Tencent is only one of many firms in China — starting from startups to telephone maker Huawei — which have rushed to announce AI merchandise this 12 months. In August, Alibaba introduced it was opening its personal AI mannequin to third-party builders.

Synthetic intelligence requires industry-specific coaching for the expertise to generate worth, Tencent’s Tong mentioned. He listed enterprise use instances in tourism, finance, public companies and customer support.

“We imagine many various clients, in reality, would profit extra by leveraging open-source fashions and use their very own enterprise information to coach for their very own fashions to fulfill the very particular wants of their industrial use instances,” he mentioned.

That designated use may also assist with information safety, he mentioned.