SHANGHAI: Serving craft beer, playing mahjong, stacking shelves and boxing, the dozens of humanoid robots at Shanghai’s World AI Conference (WAIC) this weekend were embodiments of China’s growing AI prowess and ambition.

The annual event is primed at showcasing China’s progress in the ever-evolving field of artificial intelligence, with the government aiming to position the country as a world leader on both technology and regulation as it snaps at the US’ heels.

Opening the event on Saturday, premier Li Qiang announced China would set up a new organisation for cooperation on AI governance, warning the benefits of development must be balanced with the risks.

But in the cavernous expo next door, the mood was more giddy than concerned.

“Demand is currently very strong, whether in terms of data, scenarios, model training, or artificial construction. The overall atmosphere in all these areas is very lively,” said Yang Yifan, R&D director at Transwarp, a Shanghai-based AI platform provider.

This year’s WAIC is the first since a breakthrough moment for Chinese AI this January when startup DeepSeek unveiled an AI model that performed as well as top US systems for an apparent fraction of the cost.

Organisers said the forum involved more than 800 companies, showcasing over 3,000 products – the undeniable crowd pleasers being the humanoid robots and their raft of slightly surreal party tricks.

At one booth, a robot played drums, half a beat out of time, to Queen’s “We Will Rock You” while a man in safety goggles and a security vest hyped up a giggling crowd.

Other droids, some dressed in working overalls or baseball caps, manned assembly lines, played curling with human opponents or sloppily served soft drinks from a dispenser.

While most of the machines on display were still a little jerky, the increasing sophistication year-on-year was clear to see.

The Chinese government has poured support into robotics, an area in which some experts think China might already have the upper hand over the US.

At Hangzhou-based Unitree’s stall, its G1 android – around 130cm tall, with a two-hour battery life – kicked, pivoted and punched, keeping its balance with relative fluidity as it shadowboxed around a ring.

Ahead of the conference’s opening, Unitree announced it would launch a full-size humanoid, the R1, for under US,000.

‘Digital Humans’

Most high-tech helpers don’t need hardware though.

At the expo, AI companions – in the form of middle-aged businessmen, scantily clad women and ancient warriors – waved at people from screens, asking how their day was, while other stalls ran demos allowing visitors to create their own digital avatars.

Tech giant Baidu on Saturday announced a new generation of technology for its “digital humans” – AI agents modelled on real people, which it says are “capable of thinking, making decisions, and collaborating”.

The company recently ran a six-hour e-commerce broadcast hosted by the “digital human” of a well-known streamer and another avatar.

The two agents beat the human streamer’s debut sales in some categories, Baidu said.

Over ten thousand businesses are using the technology already, the department’s head Wu Chenxia told AFP.

Asked about the impact on jobs – one of the major concerns raised around widespread AI adoption – Wu insisted that AI was a tool that should be used to improve quality and save time and effort, which still required human input.

For now, few visitors to the WAIC expo seemed worried about the potential ramifications of the back-flipping dog robots they were excitedly watching.

“When it comes to China’s AI development, we have a comparatively good foundation of data and also a wealth of application scenarios,” said Transwarp’s Yang.

“There are many more opportunities for experimentation.”