Humanoid Robots Shatter Speed Records, But Intelligence Lags, Expert Says

The World Humanoid Robot Games have showcased remarkable advancements in robotics, with humanoid robots now exceeding human speed records. However, experts like Christian Hubicki, a robotics professor at Florida State University and a former contestant on the reality show "Survivor," emphasize that despite these athletic achievements, the robots' intelligence and ability to react reliably remain limited.

During the event, a Chinese humanoid robot named Tiangong Ultra, developed by the Beijing Humanoid Robot Innovation Centre, set a new 100-meter dash record of 8.64 seconds. This surpasses the previous record held by human athlete Usain Bolt. Hubicki described the rapid pace of development as a "major milestone" for the field of legged robotics, noting that motors and joints have long possessed the capacity to outperform human physical capabilities.

"They wanted to break the record, and they're willing to break the robots to do it," Hubicki said, observing the dramatic displays of robots sparking and losing limbs in their pursuit of speed. The focus of the games, he explained, was to push the limits of how fast robots can move, with less emphasis on how they stop.

Despite the impressive speed, Hubicki pointed out that the robots' intelligence is highly specialized and lacks the nuanced understanding required for real-world applications. He explained that the robots are essentially executing pre-programmed sequences to maintain balance and forward momentum, rather than making complex, independent decisions. This is evident in other events, such as robot tennis, where external cameras are needed to track the ball.

"The higher level intelligence, we would say, is still very limited," Hubicki stated. He contrasted the robots' operational intelligence with that of advanced AI like chatbots, noting that the robots' systems are much simpler, focused on immediate tasks like calculating pose and leg placement to maintain movement.

The games featured over 2,000 robots from 666 teams competing in 51 disciplines, ranging from athletic contests to simulated tasks in factory, restaurant, and emergency settings. While the athletic performances capture headlines, Hubicki stressed the critical need for robots to develop practical intelligence to be useful and safe in human environments.

"The challenge is to get them smarter in a way that is really practical for us," Hubicki said. He raised concerns about the potential dangers of deploying robots with limited understanding of their surroundings, citing an instance where a humanoid robot struck a child because it did not perceive the child's presence. This highlights the gap between the robots' current capabilities and the reliability needed for seamless human interaction.

The discussion around robot intelligence also touches upon broader AI concerns, particularly in military applications. Hubicki noted that while humanoid robots garner significant attention, more practical and potentially concerning uses of AI and robotics are already in play, such as the use of AI-guided drones in conflict zones. The development of advanced AI, he warned, requires careful consideration to ensure it aligns with human understanding of reality and safety.