Race for the future

Unitree's Wang said his company is continuing to bring its robots into real-world settings. The robots have already been deployed in automobile factories, while the company's AI team is testing their capabilities in home settings. But he acknowledged that the company has not yet launched large-scale deployment because the robots still fall short in overall efficiency and versatility.

Wang said the "ChatGPT moment" for embodied intelligence can be measured by a clear benchmark: robots should be able to successfully complete about 80 percent of tasks in 80 percent of unfamiliar settings based on voice or text instructions.

The biggest challenge now, Wang said, is aligning the inputs and outputs of AI models with physical robots.

Xu Li, chairman and CEO of Chinese artificial intelligence company SenseTime, said: "Technological breakthroughs and real-world applications rarely happen overnight. Instead, they are built through continuous progress over time."

According to Xu, another important turning point is approaching as world models bring robots a deeper understanding of the physical world.

"Definitions and interpretations of world models may vary, but their core lies in learning the rules of the real world and using them to reason and complete long-horizon tasks in physical environments," Xu noted.

He said the true value of AI is not to replace people, but to enhance human capabilities, expand what individuals can achieve and create more development opportunities.