An interview with a Baidu robotaxi safety operator, who sits in the passenger seat and monitors how self-driving cars cope with Beijing's streets
Zeyi Yang / MIT Technology Review : Tweets: @hkanji , @baidu_inc , and @zeyiyang Tweets: Hussein Kanji / @hkanji : Robotaxi safety operator is an occupation that exists only in our time, the result of an evolving technology that's advanced enough to get rid of a driver but not good enough to convince authorities that they can do away with human intervention altogether https://www.technologyreview.com/ ... @baidu_inc : “I instinctively went for the passenger seat...and I would expect [my] car to brake by itself. And then I realized, that's not right!” Liu Yang is a Robotaxi safety driver. Follow @techreview's @ZeyiYang for a day in the life of this futuristic profession: https://www.technologyreview.com/ ... Zeyi Yang / @zeyiyang : Liu Yang has one of the strangest jobs around: to sit in the passenger seat and monitor how Baidu's self-driving cars cope with Beijing's streets. I asked him what he thinks about his job in this Q&A for @techreview https://www.technologyreview.com/ ...
Context & Ripple Effects
This interview lands mid-arc in China's robotaxi buildout: six months earlier, TechCrunch surveyed the seven Chinese operators — AutoX, Baidu, Deeproute.ai, Didi, Momenta, Pony.ai, WeRide — all racing through 2021 pilots. By 2024 that had become 16+ cities letting 19+ companies test on public roads, with Wuhan's relaxed regulatory approach making it Baidu's biggest proving ground while Beijing stayed comparatively cautious.
The piece also documents an occupation the rest of the corpus keeps confirming: two years later Amazon's [[a:874319|Zoox runs a Foster City command center with about 36 technicians assisting robot taxis remotely]], and by late 2025 Waymo pays gig workers via its Honk app to close doors and tow stranded cars. The passenger-seat monitor is the earliest form of a human-support layer that scales alongside autonomy rather than disappearing with the driver.
First-order effects
- Liu Yang's job exists because Beijing's cautious permitting requires a human aboard even where the car drives itself — every Baidu ride in the capital carries this overhead cost per trip.
Second-order effects
- As fleets expand across Chinese cities, the same dynamic spawns adjacent labor markets abroad: Zoox staffs a full remote-assistance command center and Waymo outsources physical rescues to app-based gig workers, turning 'robotaxi support' into a real employment category.
Third-order effects
- If regulators keep conditioning permits on human intervention capability, autonomy economics split geographically — fully driverless operation concentrates where rules relax (Wuhan) while monitored service persists elsewhere, and the industry's cost structure stays tied to human oversight headcount.
The trend: Robotaxis are scaling faster than regulators' willingness to remove humans, creating a durable parallel workforce of onboard monitors, remote assistants, and street-level rescuers.