'Momen ChatGPT' for Physical AI Still Needs 5 Years

Zhang Yaqin expressed his views at the 2026 Philanthropy Asia Summit. According to him, the "ChatGPT moment" for physical AI, including robots and autonomous vehicles, is still about five years away. (Photo: Yicai Global)

JAKARTA - A major breakthrough for physical AI will not come tomorrow morning. Zhang Yaqin, professor of artificial intelligence at Tsinghua University, estimates that this technology still needs about five years to reach the "ChatGPT moment".

According to a report by Yicai Global quoted on Monday, June 8, Zhang said the number of robots in the world could one day even surpass the human population. However, the road there is not short. AI must first come out of the screen and actually work in the real world.

Physical AI or physical AI is artificial intelligence embedded in moving objects, such as robots, drones, autonomous cars, and critical infrastructure. Unlike digital AI that lives in applications, chatbots, or software systems, physical AI must read the space, move, make decisions, and face a changing environment.

Zhang, who is also Dean of the Institute for AI Industry Research at Tsinghua University, said autonomous vehicles have the potential to be the first commercial application of physical AI. The reason is that the underlying technology is more mature.

The next stage, said Zhang, is large-scale commercial operations, accompanied by supporting industry policies and regulations.

However, household robots still face a more difficult path. The house is not a factory. The layout is different, the habits change, and the work is often not neat on paper. Therefore, the commercialization of household robots is considered much more difficult.

Zhang divides AI into three groups, namely digital AI, physical AI, and biological AI. According to Zhang, the current use of AI is still heavily based on digital AI and has not really entered the physical world.

When physical AI matures, the technology will be present in robots, drones, driverless cars, and critical infrastructure. At that point, the scale of deployment and the industry's influence will skyrocket.

For biological AI, the time is even longer. Research in this field includes brain-computer interfaces, i.e. technology that connects brain signals with computer devices, as well as the development of new drugs. The focus is on building basic models for cells and proteins, with the ultimate goal of a common model for biological sciences.

Zhang estimates that physical AI and biological AI can grow rapidly after the maturity of its industry infrastructure. The impact is not only technological, but also economic and social.

Regarding the expansion of Chinese AI companies abroad, Zhang reminded the importance of the principle of AI for Good, namely the use of AI for good. He said companies needed to prioritize common global problems, such as health crises, global warming, and population aging.

"Companies must prioritize solving common global challenges, including public health crises, global warming, and population aging," Zhang said.

Yicai Global also reported that Zhang highlighted three main risks of AI, namely technical failure, misuse of technology, and social impact due to massive disruption.

Technical risks can be dangerous if AI systems or robots gain access to critical infrastructure, such as financial networks, nuclear power plants, or large power grids.

Zhang said Chinese AI companies must abide by local laws and cultures when entering overseas markets. Expansion, he said, is not just about selling the same products to other countries.

According to Zhang, localization, cooperation with local companies, and respect for local rules are important requirements. In the health sector, for example, cross-country medical data privacy rules are very strict. Therefore, AI companies cannot just rely on their own algorithms without working with local hospitals.