Boson Dynamic Founder Dr. Marc Raibert and Alibaba Cloud founder Dr. Wang Jian are attending a panel discussion moderated by renowned journalist Craig Smith.
AsianFin – Marc Raibert, the legendary figure in the global robotics industry, has noted that reinforcement learning, a major sublet of artificial intelligence, is on the horizon for empowering robots during the T-EDGE Conference organized by TMTPost Group.
Raibert, who is a source of inspiration for young Chinese technologists, pointed out traditional controls, including servomechanism and vision systems, are still playing a large role in robotics although reinforcement learning is “right on the horizon.”
The role model for young robotics researchers in China is moderately optimistic about the use of artificial intelligence in the near future. “The jury’s still out,” he said during a panel discussion on AI and robotics, in response to some people’s perception that the robots will be all controlled based on self-learning very soon.
"It's true that at some place out in the future, probably reinforcement learning will take over,” Raibert acknowledged. But he went to note that physical intelligence will continue to play a major role on the level of products in the next few years, or even in five or seven years.
The founder of Boston Dynamic AI Institute said that both approaches are adopted in his institute. “We do have people working on the more extreme end-to-end data-driven approaches, but we also have people working on theory and ‘traditional stuff’,” Raibert told the audience of the T-EDGE Conference held from December 6 through December 7 in Beijing.
Raibert used the term “traditional” to refer to mainstream technologies that emerged about two years ago, just as Craig Smith, the former New York Times journalist and moderator of the panel, used the term in his question, although the former MIT professor and life-long roboticist jokingly disputed the use of the term.
The former MIT professor said that both the hardware and software of robotics are critical. He raised concerns about Silicon Valley people’s too much emphasis on software but not enough attention to hardware.
"It is true that software enables hardware beyond its basic capabilities. Eventually, the hardware is going to become the bottleneck again and the pendulum will swing back,” said Raibert.
Wang Jian, the Chief Technology Officer of Alibaba Group and founder of Alibaba Cloud, shared his view on the robotics industry in China during the panel discussion. He said that there were about 50 robotics companies in China, citing Unitree as a good example, although it is hard to determine the total number.
Wang viewed China as not just a big market but also a good place for testing new products as the country is big enough for the products made with different technological approaches.
He mentioned the massive engineering capability and massive production capacity in various sectors in China, including the robotics industry.
Regarding risks arising from the development of AI, both Reibart and Wang said that risks should not be overrated and the public should balance the risks and the potential benefits.
Reibart gave an example in San Francisco, where the human-driven car was more responsible for a road accident but the self-driving car was restricted in the wake of the accident.
Wang cited the Manhattan Declaration, signed in October 2024 by 21 AI scientists from developing countries in New York, as an effort to raise awareness about the safety, ethics and social impact of the transformative technology. The document was drafted on the sideline of the United Nations meeting “Towards a Common Understanding of AI Capabilities, Opportunities, and Risks: Forging the Path for a Positive Future for All.”
Regarding use cases for robots, Raibert thought the factory is the next place after the warehouse, which already uses many robots. He also pointed out the house is the final destination for robots because of the complex environment.
Wang predicted that some form of robots, like a companion robot that sits on the counch, may appear in the home in 10 years. Raibert thought that’s likely but that form of robots would not be all-purpose.
The panel discussion ended on a high note when Smith proposed another panel discussion at the T-EDGE Conference in 10 years to see how the robotics industry unfolds.
Comments