Humanoid Robot Sets New Half-Marathon Record in China

Humanoid Robot Sets New Half-Marathon Record in China

Over the weekend in China, a humanoid robot broke the world half-marathon record—the human record—by seven minutes.

The standout was a robot created by the Chinese firm Honor (the smartphone manufacturer), which completed the 13.1-mile race in 50 minutes, 26 seconds. The human benchmark, established by Ugandan Olympic medalist Jacob Kiplimo, is 57 minutes, 20 seconds. This result signifies a remarkable achievement especially considering that, merely a year prior, the fastest robot at the same half-marathon event took two and a half hours to finish the distance.

However, Honor’s robot was not the sole contender. The competition featured over 100 humanoid robots from 76 organizations throughout China. The robots lined up alongside 12,000 human runners in Beijing’s E-Town, albeit on separate routes to prevent mishaps. The disparity in performance between humans and robots was clearly evident.

Run, Robot, Run

A humanoid robot is crafted to replicate the structure and movement of the human body, equipped with legs, arms, and sensors that enable it to interact with its surroundings. In this instance, the victorious robot integrated features inspired by elite athletes: long legs (nearly a meter), sophisticated balance systems, and a liquid cooling system, akin to that of smartphones, to avert overheating during the race.

Furthermore, many participating robots operated autonomously, functioning without direct human intervention. Utilizing artificial intelligence algorithms, they could modify their speed, maintain equilibrium, and adapt to the terrain fluidly. Notably, the Honor robot that reached the 50-minute mark functioned autonomously. The Chinese manufacturer also showcased another robot, controlled remotely, that covered the same distance in even less time: 48 minutes, 19 seconds.

As anticipated, there were some mishaps during the race. Some robots toppled over, others strayed off course, and several required technical support along the route. While the physical capabilities of humanoid robots have progressed swiftly, their reliability is still under development. Naturally, the laughter and mockery are no longer as prevalent as they once were, replaced by applause and expressions of astonishment.

Robot Superiority

Similar to the robots that gained attention for their impressive martial arts performance a few weeks ago, this long-distance race forms part of China’s broader strategy to showcase its leadership in advanced robot development.

One doesn’t need to be a robotics specialist to recognize that this achievement illustrates that machines can surpass humans in specific physical tasks under controlled circumstances. (It’s difficult to envision the winning robot achieving the same outcome, for instance, if it started to rain during the race.) Yet humans still possess advantages: Running straight is fundamentally different from executing intricate real-world tasks, such as manipulating fragile items or engaging socially.

Nevertheless, it’s understandable that the image of a robot crossing the finish line in record time, outpacing human athletes, prompts several questions. Is this the dawn of a new era where machines redefine physical boundaries?

One might contend that a car is a machine, and those have always outpaced humans. Yet a humanoid robot is specifically designed to imitate humans. It’s more concerning to witness one best humanity at its own game—even if many of them are still stumbling along the way.

This story originally appeared in WIRED en Español and has been translated from Spanish.

Robotaxi Service Interruption Leaves Passengers Stuck on China's Roadways

Robotaxi Service Interruption Leaves Passengers Stuck on China’s Roadways

An unidentified technical issue led to several robotaxis operated by the Chinese technology company Baidu to become immobilized in traffic on Tuesday, leaving some passengers stranded for more than an hour. In Wuhan, where Baidu has rolled out many Apollo Go autonomous taxis, users on Chinese social media reported observing the vehicles malfunctioning and halting. Images and videos capture the stationary Baidu cars on congested roadways, frequently in the fast lane.

A university student in Wuhan mentioned being trapped with friends in a Baidu robotaxi for roughly 90 minutes. The vehicle malfunctioned and stopped several times before halting at an intersection. The screen in the car instructed passengers to stay seated and wait for a company representative, but after 30 minutes, they were unable to connect with customer support. After another hour of waiting, the passengers chose to exit the vehicle on their own.

Other riders also voiced their frustrations on social media regarding the inability to reach customer support. One user shared that she attempted all available help options on the app, but the phone line was inoperative, and the SOS button was not an option. She ultimately forced the door open and stepped out as traffic came to a standstill behind her robotaxi.

Baidu has yet to provide a statement. Police in Wuhan indicated that the problem might be linked to a system failure, though the situation is currently being investigated. No injuries have been reported, and it is said that passengers managed to leave the vehicles. The total number of affected robotaxis remains uncertain.

A dash cam recording depicts a vehicle going around 16 Apollo Go units that were parked on the road for over 90 minutes, with the driver narrowly avoiding them. In a separate event, a man alleged that he crashed into a halted Baidu vehicle after the car ahead of him swerved to dodge it. His orange SUV was documented with significant damage after being towed away.

AI Research Is Becoming More and More Interwoven With Geopolitics

AI Research Is Becoming More and More Interwoven With Geopolitics

The prominent AI research event, NeurIPS, recently encountered a controversy intertwining geopolitics with worldwide scientific cooperation. The conference initially set forth but quickly revoked new limitations on international participation after Chinese AI researchers indicated a possible boycott. Paul Triolo from DGA-Albright Stonebridge emphasized the necessity of drawing in Chinese researchers to serve US interests, amidst American demands to sever AI collaborations. These strains pose a risk of discouraging Chinese researchers from interacting with US academic institutions and technology firms. NeurIPS first delineated restrictions in its handbook, associating with US-sanctioned organizations, impacting researchers at companies such as Tencent and Huawei. The listing also encompassed Russian, Iranian, and other foreign entities. In light of the outcry, NeurIPS narrowed the restrictions to a list aimed at terrorist and criminal groups. The organizers acknowledged a mistake, attributing it to a lapse in communication with their legal team. The revised rule elicited a prompt reaction, especially from China, a key hub for AI expertise. Chinese academic organizations condemned the action, dissuading participation and advocating for national conferences. The China Association of Science and Technology (CAST), a notable body, revealed it would shift funding from NeurIPS attendance to endorse other conferences valuing Chinese scholars, also indicating that NeurIPS 2026 publications would not be counted towards research funding assessments, unless any policy change occurs.