At a recent hearing in the U.S. Senate, self-driving leader Waymo once again debunked the myth of "complete automation": when its self-driving taxis encounter unusual situations on the road, the vehicles will be switched to remote human drivers. In addition to being in the United States, a considerable number of these drivers come from overseas regions such as the Philippines.
Waymo Chief Safety Officer Mauricio Peña said at the hearing that the company will use remote drivers under special road conditions to ensure the smooth escape of vehicles. These operators are mostly contract workers and are located in many countries including the Philippines. This once again confirms a trend: many artificial intelligence systems packaged as "fully automated" actually rely heavily on low-paid human support in the background, and manual intervention is far more frequent than the public imagines.
In the field of autonomous driving, this kind of "manpower saving" is not an isolated case. Tesla’s Robotaxi fleet still requires human monitors to sit in the car and be ready to take over driving, while Waymo solves complex scenarios through remote takeover. Neither company has truly achieved fully unmanned operations. On the other side of the AI industry, since OpenAI led to a new round of investment boom, a large number of contract workers have been hired to label data, clean content, and train large models. The hourly wages are often only $15 and lack of benefits. This invisible labor system supports seemingly "smart" chatbots and automated systems.
Similar situations are common with other "automated" products. Presto Automation, a fast-food drive-thru system billed as "automated ordering," has order processing largely monitored and intervened by remote workers in the Philippines. Amazon once claimed that customers can "grab and go" with its Just Walk Out technology, which does not require a checkout. Behind the scenes, a large number of manpower in India is responsible for monitoring customer behavior and determining purchase details. This is quite different from the public understanding of "pure algorithmic settlement".
In the broader field of robotics and humanoid robots, the boundary between humans and machines is also far from disappearing. When Tesla showed off its robots at the "We, Robot" launch event in late 2024, it admitted that it still relied on human operators. Since then, a video of a robot falling simultaneously when a remote operator took off his headset during a demonstration went viral on social media, more intuitively exposing the close connection between so-called “autonomous robots” and remote control.
However, when the senators questioned Peña, the focus was not just on "whether anyone was involved," but more on "where are these people." Massachusetts Senator Ed Markey criticized Waymo's extensive use of foreign remote drivers as "completely unacceptable," believing that on the one hand, network delays across half the world may cause safety hazards, and on the other hand, a large number of jobs involving critical transportation infrastructure are outsourced overseas, raising concerns about national security and industrial security.
In addition to workforce distribution, lawmakers have also questioned Waymo's ties to countries such as China. Unlike Tesla, which uses its own models, Waymo's self-driving platform uses vehicles from multiple countries, including those made in China, sparking speculation about whether Alphabet's Waymo is using this to circumvent import restrictions on Chinese cars. In response to questions, Peña emphasized that Waymo’s self-driving system is installed in the United States, trying to downplay concerns about data and security risks that may arise from “connected Chinese vehicles driving on American roads.”
From remotely taking over drivers to overseas contract workers, to the use of Chinese-made vehicles, this hearing focused on a core contradiction of the current autonomous driving and AI industries: on one side is the marketing narrative of "fully automated" and "unmanned", and on the other side is the highly globalized human and supply chain network silently operating behind the scenes. For ordinary users, hailing a "self-driving taxi" may seem like dealing only with algorithms, but in remote data centers, there are often pairs of human hands ready to take over the steering wheel when the system fails.