When a Waymo robotaxi in Los Angeles refuses to budge because a door jammed or the car ran out of charge, the rescue doesn't always look like the cinematic image of engineers in white coats. More often it looks like a tow truck rolling up and a driver asked to do something ordinary: close a door or move the vehicle, then get paid — sometimes as little as $22.
What sounds like a punchline is a snapshot of how automation and gig work are colliding on public streets. Recent reporting in the Washington Post, and a follow-up in Gizmodo, found tow operators using the Honk dispatch app to accept tiny, on-demand jobs from companies operating robotaxis. Tow operators such as Evangelica Cuevas described offers of "$22 to $24" to close stuck doors and "$60 to $80" to tow a disabled robotaxi — gigs that can feel like patchwork insurance for unreliable software and hardware.
A gig-economy safety net
These are not the high-tech fixes complex startups promised a decade ago. They are an ad hoc human layer propping up automated fleets. Cuevas and other tow-company owners told reporters they sometimes take the work not because it's lucrative but because it's steady and immediate. For drivers already chasing short, unpredictable gigs, an extra $22 to close a door — or a larger fee to tow — can matter.
Georgios Petropoulos, a data scientist at the University of Southern California, told the Post that humans remain essential to keep automated services running safely and efficiently. That human-in-the-loop reality is hardly a new concept in robotics, but seeing it play out as pay-per-incident towing and door closure gigs puts a sharp, human face on the tradeoffs of automation.
There are two threads tangled here. One is technical: self-driving systems are better every year, but edge cases — hardware faults, battery depletion, sensor confusion around curbside parking — still require human judgment or muscle. The other is economic: companies often route the fix through local gig platforms instead of building full-time, on-the-ground maintenance teams.
The local picture in Los Angeles is worth noting because it’s both a testing ground for robotaxis and a city with an entrenched gig-and-tow economy. Drivers who pick up Honk jobs describe unpredictability and low margins; they also describe a steady drumbeat of these small jobs as robotaxis multiply.
What this says about automation — and what it doesn't
There’s a darker pattern beneath the anecdote. Automation often promises to eliminate the need for messy human labor, but in practice it can shift work into more precarious forms. A stuck robotaxi doesn't create a driving job so much as it creates a microtask: someone to come fix a product the automated system couldn’t finish. That shift raises questions about wages, labor protections, and who bears the costs when technology falls short.
It also complicates the public conversation about how quickly and widely to deploy autonomous vehicles. City planners and regulators weigh safety, congestion, and environmental promises against real-world failures. Meanwhile, companies are deploying features like conversational navigation and advanced AI copilots to smooth customer experience; see how broader navigation AI is evolving in projects such as Google Maps getting a Gemini AI copilot for navigation. Those advances matter, but they don't erase the need for pavement-level fixes.
Nor does this anecdote answer the larger, philosophical debate about where AI is heading. Technologists and skeptics alike are arguing over whether we're approaching human-level intelligence and what that would mean for work and society — a debate playing out in both labs and tow yards (AI’s tipping point: pioneers vs. skeptics).
There are also practical safety questions. Who is accountable when a gig worker is asked to interact with a vehicle full of passengers and automated controls? Are companies training or insuring these on-demand helpers? Reports so far suggest the model is more stopgap than strategy: quick fixes routed through existing informal labor networks rather than a promise of long-term jobs or official partnerships.
The image is oddly domestic: a tow truck at dusk, a lone worker clicking a smartphone, the hum of a city never fully that impressed by the future. Robotaxis will get better. The surrounding economy will adapt in ways we've already started to recognize — new gigs, new forms of precarity, and new questions about who pays when machines fail.
For now, when an ostensibly driverless car needs help, it often gets a human. Sometimes that human is in a call center miles away; sometimes they are in a tow truck getting $22 to close a door.