![DoorDashers Closing Waymo Self-Driving Car Doors: The Future of Gig Work [2025]](https://tryrunable.com/blog/doordashers-closing-waymo-self-driving-car-doors-the-future-/image-1-1770988338187.jpg)
Door Dashers Closing Waymo Self-Driving Car Doors: The Future of Gig Work [2025]
Last year, the internet discovered something that seemed almost too absurd to be real. Door Dash drivers were getting paid to close car doors. Not their own cars. Not as part of a prank. But as an actual, paid job through a pilot program between Door Dash and Waymo, the self-driving car subsidiary of Alphabet.
The task was straightforward: if a passenger left a Waymo autonomous vehicle door slightly ajar, nearby Door Dash drivers would receive a notification offering them
What makes this story worth understanding isn't just the novelty of paying humans to do what a machine could theoretically do. It's what the program reveals about the actual limits of self-driving technology, the economics of autonomous delivery, and how companies are creatively filling the gaps between what their AI can do and what the real world demands. This isn't a failure of Waymo's technology. It's a pragmatic solution to a problem that automation alone couldn't solve, and it's reshaping how we think about the future of work in an AI-driven economy.
The Waymo-Door Dash partnership started delivering food in Phoenix, Arizona in October 2025. Then, quietly, in early 2025, the door-closing pilot launched in Atlanta. Most people didn't notice. But the gig workers did. And the story exploded across social media, forcing conversations about automation, labor economics, and whether the future of work is really about replacing humans or integrating them differently into systems where AI handles the complex problems and humans handle the exceptions.
Let's dig into what's actually happening here, why it matters, and what it means for the thousands of gig workers and the millions of people relying on autonomous delivery services.
TL; DR
- **Waymo pays Door Dashers 5 bonus to close autonomous vehicle doors left ajar by passengers in Atlanta pilot
- Self-driving cars can't depart if doors aren't fully closed, making this a genuine operational constraint, not a design flaw
- Gig workers are becoming exception handlers for automation systems, a growing trend across logistics, delivery, and warehouse work
- Economic arbitrage at scale: Waymo saves money by using distributed gig workers instead of dedicated fleet staff
- The future isn't human replacement, it's human integration into automated workflows where AI handles routine tasks and humans handle edge cases
The door-closing task offers an effective wage of
The Unexpected Origins: How a Simple Problem Revealed the Limits of Automation
Waymo's autonomous vehicles are extraordinarily sophisticated. They use a combination of LiDAR, radar, and computer vision to navigate complex urban environments, interpret traffic signals, predict pedestrian behavior, and handle dozens of real-time decisions every second. The cars can merge onto highways, navigate construction zones, and make split-second emergency maneuvers. What they can't do, apparently, is ensure passengers close their doors completely.
This isn't a technical oversight. It's a safety design choice. Waymo vehicles are optimized to maximize passenger safety, which means allowing passengers to exit without forcing doors shut behind them. The vehicle can detect whether a door is fully closed through magnetic sensor feedback. If it isn't, the car won't move. The system works perfectly from a safety perspective. The problem is operational efficiency.
When a passenger exits a Waymo in downtown Atlanta and accidentally leaves the door partially open, the autonomous system detects it and prevents the vehicle from proceeding to the next passenger pickup or delivery location. The car sits immobilized, unable to complete its assigned route, creating a queue of delayed pickups and deliveries. This creates a cascade of problems downstream. Other passengers are waiting. Other food orders are cooling. The vehicle's utilization rate drops.
There are only so many solutions to this problem. Waymo could send a dedicated employee to every location where doors are left open, but that defeats the purpose of autonomous delivery entirely. They could redesign the vehicle to automatically close doors, but that introduces safety risks and liability concerns. They could implement an app notification system where passengers close the door remotely, but that requires passenger adoption and cooperation. Or they could distribute the task across a vast network of people already in the area performing delivery work.
Waymo chose the last option.
What's brilliant about this approach is that it leverages an existing distributed workforce that's already geographically scattered and equipped with mobile devices. Door Dash drivers are already throughout Atlanta making deliveries. They're already navigating the city, already checking their phones, already accustomed to receiving gig work offers. The marginal cost of adding a door-closing task to a Door Dasher's queue is nearly zero from Waymo's perspective. From the driver's perspective, it's a quick task that takes maybe 30 seconds and pays $6.25.
The economics almost make sense. The question is whether this is a clever operational solution or a preview of a more unsettling future where automation is used to centralize profits while distributing labor across an increasingly fragmented gig workforce.
Waymo door-closing tasks offer significantly higher hourly earnings (
Understanding the Waymo-Door Dash Partnership: A Symbiotic Relationship Built on Trust
Waymo and Door Dash didn't wake up one morning and decide to pay Door Dashers to close car doors. This pilot program is the result of a deeper partnership between the two companies that started years ago.
In 2020, Waymo began working with Door Dash to explore autonomous delivery possibilities. Both companies recognized a mutual opportunity. Door Dash was facing driver shortage issues, rising labor costs, and increasing pressure to improve delivery speeds. Waymo had autonomous vehicles proven to work reliably in urban environments but limited use cases beyond ride-hailing. By combining forces, Door Dash could offer autonomous delivery as an option for customers while reducing pressure on its driver workforce for long-distance deliveries. Waymo could generate revenue from its autonomous fleet and prove real-world commercial value.
The partnership deepened when both companies expanded beyond their home markets. Waymo already had autonomous fleets operating in Phoenix, San Francisco, and Los Angeles. Door Dash's presence in these cities meant there was immediate opportunity to test integrated delivery. In October 2025, the companies officially launched autonomous delivery in Phoenix. Customers could select "opt in to autonomous delivery" during checkout. If a Waymo was available in their area, their food would arrive in an autonomous vehicle. Customers would retrieve their orders directly from the trunk.
This is where the door-closing pilot fits into the bigger picture. As Waymo and Door Dash scaled autonomous delivery to Atlanta and started operating in denser urban environments, they discovered that passenger-initiated vehicle door issues became more frequent. The pilot program wasn't a scrambled response to a crisis. It was a calculated solution to maintain the efficiency targets both companies promised to investors.
From Door Dash's perspective, this partnership is crucial. Autonomous delivery reduces the demand for drivers long-term, which sounds bad for worker income. But it also solves logistics problems that current drivers face, like excessive traffic during peak hours or dangerous driving conditions. For drivers themselves, the partnership offers flexibility. The door-closing task is optional. Drivers can accept it or ignore it based on convenience. Most importantly, it's additional income above and beyond their normal delivery rates.
From Waymo's perspective, outsourcing the door-closing task to Door Dash drivers eliminates the need for dedicated fleet maintenance staff positioned throughout Atlanta. Instead of hiring 15 full-time employees to handle exceptional cases like open doors, Waymo distributes the responsibility across an already-existing network of thousands of drivers.
The Economics of the Door-Closing Task: How Gig Work Arbitrage Works at Scale
Let's break down the actual economics. Door Dashers get paid
For comparison, the median Door Dash driver makes between
However, there's a crucial asterisk. Drivers don't get
From Waymo's perspective, the math is compelling. Imagine Waymo needs to maintain 15 full-time field staff throughout Atlanta to handle exceptional cases like open doors, stuck sensors, or minor mechanical issues. That's
Alternatively, Waymo can distribute this responsibility across Door Dash's network of thousands of drivers. Even if door-closing tasks occur 100 times per day across the fleet, that's only
This is what economists call labor arbitrage through platform economics. Waymo isn't creating a new job category. It's distributing work across an existing platform that has the infrastructure, payment systems, and worker access already built. The marginal cost of adding another task type to the Door Dash app is nearly zero. The benefit is substantial operational efficiency for Waymo.
For gig workers, this has mixed implications. On one hand, additional income opportunities are always welcome. On the other hand, it reveals how completely fragmented and task-based gig work is becoming. Door Dashers aren't door-closers who occasionally deliver food. They're becoming workers who might close doors, deliver food, shop for groceries, or complete a dozen other microtasks based on what algorithms deem profitable in their geographic area at any given moment.
Estimated data shows a balanced distribution among potential solutions, with app notifications slightly favored due to ease of implementation and user engagement.
The Real Problem Waymo Solved: Autonomous Vehicles Aren't Actually Autonomous
Here's the uncomfortable truth about autonomous vehicles that most discussions skip over. They aren't actually autonomous. They're highly automated within defined parameters. When everything works as designed, they operate without human intervention. But when exceptions occur—and exceptions always occur in complex real-world environments—the system either fails or requires human intervention.
Waymo's vehicles can drive themselves under most conditions. They can't, however, handle exceptions that fall outside their programmed parameters. A door left partially open isn't something the vehicle's AI training equipped it to handle. The vehicle can detect the problem (through magnetic sensors). It can communicate the problem (to Waymo's ops center or to nearby workers). But it can't solve it independently in a way that respects passenger safety and dignity.
This is the core insight: automation and autonomy aren't the same thing. Waymo's vehicles are highly automated, but they're not truly autonomous because they require human fallback systems for exception handling.
And Waymo is far from alone in discovering this limitation. Self-driving truck companies like Waymo's UPS partnership, Amazon's delivery drones, and even warehouse automation systems all face the same reality. Automated systems handle the high-volume, routine work efficiently. But exceptional cases—the 5% of situations that don't fit the standard script—still require humans.
Waymo's solution is elegant because it acknowledges this truth directly. Rather than overpromising full autonomy and underdelivering, the company designed a system that leverages automation for what it does well (driving) and human intervention for what it doesn't (handling exceptions). The door-closing task is small enough that it's economically viable to distribute across gig workers rather than employ dedicated staff.
What's interesting is how this model will evolve. As Waymo's vehicles operate across more cities and handle more delivery scenarios, they'll accumulate data about which exception types occur most frequently. They'll likely implement technical solutions to reduce exceptions. Perhaps future vehicles will have doors that close more predictably, or tactile feedback that prevents passengers from leaving them ajar. But until then, and for countless other exception types we haven't even thought of yet, humans are filling the gaps.
The Gig Worker Perspective: Fragmentation and the Future of Work
For the Door Dash drivers receiving these notifications, the experience is surreal. One minute you're thinking about where to pick up your next food delivery. The next minute you get a ping offering money to close a car door. It's not a large amount. It's not consistent. But it represents something bigger about how gig work is evolving.
Traditional employment was vertically integrated. You worked for a company. That company owned your labor, determined your tasks, managed your schedule. You had benefits, stability, and a clear role. Gig work dismantled that model. Now workers are classified as independent contractors picking up tasks from an algorithmic marketplace. The tasks aren't bound to a single company or vertical. They're microtasks distributed across a network of platforms.
Door Dashers aren't just Dashy. They might also be available on Instacart, Doordash, Grubhub, and now Waymo. Each platform treats them as independent workers available for different task types. The fragmentation is so complete that a single worker might simultaneously work for four different companies, performing four different job types, with four different pay structures and zero job security across any of them.
The door-closing task epitomizes this fragmentation. It's not a delivery job. It's not a service job in any traditional sense. It's a task that exists because automation created an exception that requires human resolution. And the most efficient way to resolve it economically is to distribute it across workers who happen to be in the geographic area already doing something else.
Some gig workers see opportunity in this. Additional income sources, flexibility, the ability to earn money during slow periods. Others see exploitation. The work is still low-wage, still insecure, still lacks benefits. And now it's increasingly fragmented into absurd microtasks that defy anyone's traditional conception of a "job."
Waymo pays competitively for the door-closing task compared to Door Dash delivery. But it's still built on the same fundamental economic model: distribute work to workers classified as independent contractors with minimal protections, benefits, or job security.
The Waymo-DoorDash partnership has expanded autonomous delivery from initial testing in 2020 to five cities by 2025, showcasing their strategic growth. (Estimated data)
Technical Limitations of Self-Driving Cars: Why Doors Matter More Than You Think
People often think of autonomous vehicles as either fully working or completely broken. But the reality is far more granular. Waymo's vehicles have solved problems that humans find trivially easy but machines find extremely difficult. They can navigate chaotic intersections, interpret hand signals from traffic cops, and predict how pedestrians will move based on subtle body language.
Simultaneously, they can't handle problems that seem simple to humans. Closing a car door is one example. But there are others. What happens if someone leaves a backpack in the car? If the trunk won't close properly? If a passenger is having a medical emergency? If weather is extreme? These exceptions require human judgment, flexibility, and the ability to respond to novel situations.
The door-closing problem specifically reveals something important about how autonomous vehicles interact with humans. The vehicles are designed to be safe, which means they won't move if any door isn't fully secured. This is correct from a safety perspective. The passenger sitting in a vehicle with an open door is vulnerable. But it creates an operational constraint. If the door detection system is slightly miscalibrated, or if a passenger closes the door 95% of the way, the vehicle assumes something is wrong and stops.
Most sophisticated autonomous vehicle companies are working on solutions to reduce exceptions. Waymo is investing in better sensor calibration, more robust door mechanisms, and AI training that can predict when passenger behavior is likely to create problems. Tesla's Full Self-Driving beta implementation handles some edge cases through continuous learning from millions of driving hours. But these are slow, expensive, long-term solutions.
The door-closing pilot is Waymo's pragmatic near-term answer. Acknowledge the limitation. Distribute the exception handling. Continue improving the core technology. It's honest systems design.
The Broader Trend: Exception Handling as the New Gig Economy
Waymo's door-closing task isn't an isolated phenomenon. It's part of a broader trend where gig platforms are becoming exception-handling networks for automated systems.
Amazon's Flex drivers don't just deliver packages anymore. They also help resolve delivery exceptions like access issues to buildings, customer availability problems, or failed delivery attempts. Instacart shoppers don't just pick items from shelves—they handle substitutions, check expiration dates, and resolve inventory discrepancies that automated systems can't. Even Uber, despite heavy investment in autonomous vehicle development, continues expanding human driver availability for types of trips or locations where their AI can't operate reliably.
The pattern is consistent. Automation handles the bulk volume of routine work, often with 85-95% efficiency. But the remaining 5-15% of exceptions require human flexibility, judgment, and problem-solving. Rather than investing heavily in more AI to handle these exceptions, companies are distributing them across gig platforms.
This creates an interesting economic situation. Gig platforms have massive distribution networks of workers already equipped with devices, payment systems, and location data. Adding new task types to the platform is trivial from a technical perspective. The marginal cost of distributing an additional task type to thousands of workers is nearly zero. The workers are already available, already accepting tasks, already accustomed to the income variability.
From the platform's perspective, this is more efficient than building specialized solutions for each exception type. From the worker's perspective, it's additional income opportunities but also increasing fragmentation of job roles and responsibilities.
This trend will likely accelerate as automation becomes more sophisticated and more companies face the same reality Waymo discovered. The future isn't humans versus automation. It's humans integrated into automated systems as the exception handlers.
Estimated data shows that while automation handles 85-95% of tasks efficiently, 5-15% require human intervention for exceptions.
Atlanta as the Testbed: Why This City Matters for Autonomous Delivery
Waymo chose Atlanta for the door-closing pilot deliberately. The city represents a unique combination of factors that make it ideal for testing autonomous delivery at scale.
Atlanta has a sprawling urban environment with complex traffic patterns, diverse neighborhoods, and challenging geography. The city's road networks, traffic signal systems, and pedestrian behavior patterns are distinct from Waymo's home markets in the West. Testing in Atlanta validates whether autonomous vehicles can operate reliably in different urban contexts, not just California sunshine.
The city also has significant Door Dash market penetration. Thousands of active drivers are already distributed throughout Atlanta, making the logistics of recruiting door-closing workers straightforward. The Door Dash platform already has infrastructure in place, driver onboarding processes, and payment systems operating in Atlanta. Adding a new task type to existing infrastructure is easier than building new recruitment and payment systems from scratch.
Atlanta also has a climate that creates interesting test conditions. The city experiences humid summers, occasional ice storms, and diverse weather patterns. These conditions stress-test autonomous vehicle sensors differently than California's predictable sunshine. A car door that works perfectly in dry 72-degree weather might behave differently when humid and stuck, or when ice has formed on the hinges.
From a regulatory perspective, Atlanta and Georgia have been relatively amenable to autonomous vehicle testing. The state doesn't require special permits for autonomous vehicle operation in many cases. The city government has been supportive of autonomous delivery pilots as potential solutions to congestion and delivery problems.
The door-closing pilot in Atlanta also generates valuable data. Waymo can track exactly how frequently doors are left ajar, which types of passengers are most likely to do it, what time of day it happens most often, and whether it's increasing or decreasing as drivers become more accustomed to autonomous vehicles. This data informs future vehicle design, operational procedures, and exception-handling strategies.
Passenger Experience: Will Customers Care About Door-Closing Assistance?
From a customer perspective, the door-closing pilot is mostly invisible. Door Dash customers ordering food or packages aren't told that Waymo pays drivers to close car doors. It's an operational detail that doesn't affect the customer experience directly.
But it does affect the customer experience indirectly. When a Waymo vehicle's door is left ajar and nobody closes it, that vehicle can't complete its delivery route. Other customers waiting for their orders experience delays. Food that should arrive warm arrives cold. The autonomous delivery service appears less reliable.
By implementing the door-closing pilot, Waymo maintains reliability standards that customers depend on. Delivery times stay predictable. Service quality stays consistent. The customer sees a seamless autonomous delivery experience without understanding all the human exception handling happening behind the scenes.
This raises an interesting question about transparency. Should customers know that human workers are helping their autonomous deliveries succeed? Should gig workers be identified somehow? Should customers be able to tip door-closers separately from drivers?
Currently, none of this happens. The door-closing work is completely transparent to customers. It's background work that maintains the fiction of full autonomy while relying on human integration.
As autonomous delivery scales, this transparency question becomes more important. Customers have growing expectations about labor practices and worker treatment. If companies want public trust in autonomous systems, they might need to be more transparent about the human labor supporting them.
Estimated data suggests a potential imbalance in door-closing task distribution, with affluent areas possibly receiving a higher share of tasks. This could indicate geographic inequality in gig work availability.
Scaling Challenges: What Happens When Waymo Expands to 10 Cities?
The door-closing pilot works in Atlanta because Waymo's autonomous fleet is still relatively small and concentrated geographically. Thousands of Door Dash drivers are distributed throughout the city, making response times short. The system is efficient.
But as Waymo expands to more cities, scaling becomes challenging. Every new city means recruiting and training door-closer workers. Every expansion means building relationships with local gig platforms. Every market has different labor dynamics, compensation expectations, and worker availability.
Scaling also means the volume of door-closing tasks potentially increases. Maybe a handful of doors are left ajar daily in Atlanta. But if Waymo scales to 100 cities with thousands of vehicles in each, there might be thousands of doors left ajar daily. At that scale, distributing the work across gig platforms becomes less efficient than implementing technical solutions.
This creates pressure for Waymo to invest in preventing the exception rather than just handling it. Better door designs that are harder to leave partially open. Automated reminders to passengers before exiting. Doors that close automatically after a certain timeout period. These solutions are more expensive to implement at first, but more efficient at scale than paying thousands of workers nationally to close doors.
The door-closing pilot is probably a temporary solution that makes sense at current scale. But it's revealing how Waymo thinks about growth, tradeoffs, and the role of human labor in autonomous systems. It's pragmatic, not permanent.
Regulatory and Legal Implications: Who's Responsible If Something Goes Wrong?
The door-closing pilot creates interesting regulatory questions that haven't been fully answered yet. If a Door Dasher improperly closes a door, and then the vehicle moves and the door opens during transit, who's liable? Is it Waymo for relying on untrained workers? Is it Door Dash for not ensuring workers followed proper procedures? Is it the Door Dasher who completed the task incorrectly?
These questions matter because autonomous vehicles operate in a heavily regulated space. Insurance companies, regulators, and lawyers need clear liability structures. The door-closing pilot exists in a gray area where traditional employment law, gig worker protections, and autonomous vehicle liability all overlap.
Currently, Door Dashers completing the task are instructed to verify the door is fully closed before confirming completion. This suggests Door Dash and Waymo are trying to establish that workers performed the task correctly, shifting liability slightly toward the worker. But the worker isn't an employee of Waymo. They're not trained by Waymo. They have no employment relationship with Waymo. How much responsibility can actually be placed on them?
From a regulatory perspective, federal and state regulators are still developing frameworks for autonomous vehicle operation. The door-closing pilot doesn't squarely fit existing categories. It's not exactly ride-sharing (because there's no passenger in the vehicle). It's not exactly delivery (because the task isn't delivering anything). It's exception handling for autonomous systems, which is a new operational pattern that regulations haven't caught up with yet.
Australian regulators have been more proactive than American regulators in developing autonomous vehicle frameworks. But even they haven't specifically addressed exception handling by distributed gig workers. This is uncharted regulatory territory.
As the practice scales, expect regulatory scrutiny. Expect questions about worker training, liability structure, and safety protocols. The next few years will likely see regulatory guidance emerging around how autonomous systems can safely integrate human exception handlers.
Worker Rights and Labor Concerns: The Gig Economy's Dark Side
For all the economic analysis of the door-closing task, there's an uncomfortable labor reality underneath. Gig workers are classified as independent contractors rather than employees, which means they're excluded from virtually all labor protections. No minimum wage (unless they happen to work in places that mandate gig minimum wages). No benefits. No unemployment insurance. No workers' compensation if they're injured. No job security.
The door-closing task exemplifies gig work's worst features. It's low-wage work (when amortized), insecure, available without warning, and performed without any relationship between worker and employer. A Door Dasher might receive a door-closing notification once per week. They can't plan around it. They can't build it into their income forecasting. It's just another variable in an already unpredictable income stream.
Worker advocacy groups have raised concerns about the door-closing pilot. Some argue it shows how companies are trying to obscure human labor behind automation narratives. Others argue it shows exploitation, where gig workers are treated as disposable exception handlers. These perspectives aren't unreasonable.
Waymo and Door Dash argue the tasks are optional and supplementary, offering additional income without replacing existing work. That's technically true. But it's also an excuse that avoids deeper questions about worker protections in the gig economy.
The door-closing pilot is small enough that it won't drive major labor reforms. But it's a symbol of a larger problem. As automation creates more exception handling tasks, gig platforms will likely distribute more of them across fragmented workforces. Without regulatory intervention or labor organizing, gig workers will continue bearing the risks of platform economics.
Future Technologies: Will Humans Still Close Doors?
Looking ahead, the question isn't whether humans will close Waymo doors forever. It's how long this will remain necessary before technical solutions make it obsolete.
Waymo is likely already investing in solutions. Computer vision improvements could allow vehicles to detect whether doors are truly secure and provide better feedback to passengers. Door mechanisms could be redesigned to be harder to leave partially open. Sensors could become more sensitive and less prone to false positives. Waymo could implement a system where doors automatically close after a timeout period.
Each of these solutions costs money to implement and test at Waymo's scale. But they're probably cheaper long-term than paying gig workers to close doors across multiple cities indefinitely.
The door-closing pilot is essentially a holding pattern. Waymo discovered a problem, implemented a practical short-term solution using existing infrastructure, and is presumably investing in long-term technical solutions. By 2028 or 2029, the door-closing tasks might be completely unnecessary.
But the bigger pattern—humans handling exceptions in automated systems—will continue. As autonomous vehicles become ubiquitous, they'll encounter more edge cases. As Waymo scales, the matrix of possible exception types will expand. Technical solutions will handle some of them. But human exception handlers will likely remain necessary for the long tail of unexpected situations that AI training can't fully anticipate.
The Bigger Picture: Automation, Work, and the Future Economy
Waymo's door-closing pilot is ultimately a small story with big implications. It's a single task at a single company in a single city. But it reveals something crucial about how automation is actually reshaping work in the 2020s.
The narrative we often hear is binary. Automation either replaces jobs entirely, or it doesn't. In reality, it's far more complex. Automation is restructuring work, fragmenting it into component tasks, and distributing some of those tasks across distributed gig platforms while completely eliminating others.
Waymo didn't eliminate the need for door-closing. It fragmented that work and distributed it across gig workers who were already geographically dispersed. This is more efficient than traditional employment but more precarious than alternatives. It requires no permanent commitment from workers, no benefits, no job security. But it also offers flexibility, multiple income sources, and work that exists only when needed.
This pattern will likely expand. As companies deploy more automated systems, they'll discover more exceptions that require human handling. They'll distribute those exceptions across gig platforms rather than hiring dedicated staff. The gig economy will become increasingly fragmented, with workers piecing together income from multiple platform-distributed microtasks.
From a labor perspective, this is concerning. From an economic efficiency perspective, it's rational. From a technology perspective, it reveals the actual limitations of automation. We don't have machines that can handle all human-level tasks. We have machines that can handle specific, well-defined tasks extremely well. For everything else, we distribute the work across humans willing to do it.
The future economy will likely involve both. Highly automated systems handling routine high-volume work. Distributed human workers handling exceptions, edge cases, and problems that fall outside automation's defined parameters. The relationship between humans and machines won't be replacement. It'll be integration.
Waymo's door-closing pilot is a microcosm of this future. A sophisticated autonomous vehicle that can navigate complex urban environments, handle real-time decision-making, and execute thousands of driving decisions per second. Paired with a fragmented network of gig workers handling one of the few problems the sophisticated system can't solve alone.
This is neither the utopia of full automation nor the dystopia of complete human replacement. It's the practical future of work, already being created in real time across companies like Waymo and Door Dash.
Industry Implications: What Other Companies Are Learning
Waymo's experiment isn't happening in isolation. Other autonomous vehicle companies and automation leaders are watching closely and drawing their own conclusions.
UPS, which partnered with Waymo for package delivery, is likely studying the door-closing pilot carefully. If autonomous delivery requires similar exception handling across UPS's network, that has huge implications for their implementation strategy. Do they hire exception-handling staff? Do they distribute the work across their own employees? Do they build relationships with gig platforms like Door Dash?
Amazon, which is heavily invested in autonomous delivery drones and logistics automation, faces similar questions. Amazon's Flex drivers currently handle delivery exceptions. The company could potentially formalize a system where drivers specifically handle exceptions that Amazon's automation creates. This would be similar to Waymo's model but with employees rather than gig workers.
Tesla's Full Self-Driving development is watching because it reveals the practical challenges of scaling autonomous systems. Elon Musk has promised that Tesla's autonomous vehicles will eventually be fully autonomous with zero human intervention. The reality is proving more complex. Every deployment reveals exceptions the system can't handle. Tesla hasn't publicly addressed how it plans to handle these exceptions at scale.
For robotics companies and warehouse automation, the door-closing pilot illustrates a broader pattern. Automated systems are often 80-90% sufficient for their intended purpose. That final 10-20% of edge cases, exceptions, and novel situations requires human judgment. Rather than investing heavily in more automation to handle these cases, it might be more economical to integrate humans into the workflow.
This has significant implications for ROI calculations on automation investments. Companies might overestimate automation's autonomous capacity and underestimate the need for ongoing human integration. The future success of automation projects might depend less on the sophistication of the algorithms and more on how well they integrate humans for exception handling.
Environmental and Social Impact: The Hidden Costs of Micro-Optimization
Waymo's door-closing pilot creates some unintended environmental consequences worth considering.
If a Door Dasher is making a delivery on the opposite side of town and receives a door-closing notification, they might travel specifically to close that door. This creates additional vehicle miles traveled, additional fuel consumption, and additional emissions. Individually, one trip to close a door isn't significant. But aggregate across thousands of tasks per day across multiple cities, the environmental impact becomes meaningful.
Waymo probably accounts for this through their routing algorithms. They likely only send door-closing notifications to Door Dashers who are already in the vicinity. But we don't have full visibility into how this works. It's possible that some door-closing notifications create excess vehicle miles that wouldn't otherwise occur.
There's also a social equity angle. If door-closing tasks are concentrated in lower-income neighborhoods, it creates geographic inequality in gig work availability. Conversely, if they're concentrated in affluent areas, it creates a pattern where wealthier customers' service reliability is directly subsidized by gig workers' unpaid exception-handling labor.
Door Dash and Waymo haven't been transparent about these impacts. The door-closing pilot exists in a data black hole. How many doors are left ajar daily? How far do workers travel to close them? Which neighborhoods generate most door-closing tasks? Is compensation truly competitive, or does the
These questions matter because they determine whether the door-closing pilot is a genuine innovation or a clever way to shift costs onto workers and the environment.
What's Next: How the Door-Closing Pilot Will Evolve
The door-closing pilot is still in early stages. It's running in Atlanta with what's likely a limited number of vehicles and drivers participating. We can speculate about what comes next.
First, Waymo will likely monitor the pilot closely, tracking completion rates, response times, customer impact, and economic efficiency. If it's working well—if door-closing tasks are being completed reliably and the cost is lower than alternatives—Waymo will expand the program to other cities where they operate.
Second, Waymo is probably collecting data on when and why doors are left ajar. Understanding patterns helps predict the behavior and potentially implement technical interventions. Maybe certain vehicle models are more prone to doors being left ajar. Maybe certain times of day or customer demographics account for most instances. This data drives future vehicle design decisions.
Third, as the program scales, Waymo might develop better integration with Door Dash. Instead of one-off task notifications, it could become a formal partnership with specific drivers trained for exception handling, guaranteed compensation, or priority access to profitable tasks.
Fourth, other autonomous vehicle companies will likely launch similar programs. Cruise, Tesla, Aurora, and international competitors will discover their own versions of the door-closing problem. They'll implement their own versions of distributed exception handling.
Final stage, technical solutions will eventually reduce or eliminate the need for door-closing tasks. But by then, the infrastructure, norms, and economics of distributing exception handling across gig platforms will be established. Other exception types will fill the void.
The door-closing pilot isn't permanent. But the pattern it represents—automating routine work, distributing exception handling across gig workers, and optimizing costs through platform economics—is probably permanent.
FAQ
What is the Waymo door-closing pilot program?
The pilot program is a collaboration between Waymo and Door Dash where Door Dash drivers are paid to close the doors of Waymo autonomous vehicles in Atlanta if passengers accidentally leave them partially open. Drivers receive
How does the door-closing payment system work?
Door Dash drivers receive notifications through their app when a nearby Waymo vehicle has a door left ajar. They can choose to accept the task, travel to the vehicle, close the door properly, and verify completion through the app. Upon verification, they receive
Why does Waymo need humans to close car doors?
Waymo's autonomous vehicles include safety features that prevent the vehicle from moving if any door isn't fully closed. While this is essential for passenger safety, it creates an operational constraint. If doors are left partially open, vehicles can't proceed to next pickups or deliveries. Distributing the exception handling across existing gig worker networks is more cost-effective than employing dedicated staff.
Is this pay competitive compared to standard Door Dash delivery work?
Yes, significantly. The
What does this reveal about autonomous vehicle limitations?
It demonstrates that fully autonomous vehicles still require human integration for exception handling. The vehicles can drive themselves through complex environments but can't handle all real-world situations independently. This reveals the difference between automation (handling routine tasks reliably) and autonomy (handling all situations without human input).
Are gig workers being exploited through this program?
The program offers optional supplementary income above standard delivery rates, which isn't inherently exploitative. However, it exemplifies concerns about gig worker fragmentation, lack of job security, and absence of benefits. The $11.25 task rate is competitive only relative to poor baseline gig work compensation.
Will humans always need to close Waymo car doors?
Probably not. Waymo is likely investing in technical solutions like improved door sensors, better door mechanisms, automated reminders to passengers, or doors that close automatically after timeout periods. The pilot is likely a temporary operational solution while long-term technical fixes are developed.
What are the regulatory implications of paying workers for this task?
The pilot creates gray areas in liability, worker classification, and autonomous vehicle operations that regulators haven't fully addressed. Questions about who's responsible if doors are closed incorrectly, how workers should be trained, and how this fits existing labor law are still unanswered.
How does this expand to other companies and industries?
Other autonomous vehicle companies (Cruise, Tesla, Aurora) will likely discover similar problems requiring human exception handling. More broadly, companies using automation across logistics, warehousing, and delivery will probably distribute similar exception-handling tasks across gig platforms.
What does the door-closing pilot mean for the future of work?
It suggests the future economy will integrate humans and machines, with automation handling high-volume routine work and humans handling exceptions and edge cases. Rather than full replacement, we'll see fragmented, distributed gig workers integrated into automated systems as exception handlers.
Conclusion: A Glimpse Into an Integrated Future
Waymo's decision to pay Door Dash drivers to close car doors isn't a footnote in the autonomous vehicle story. It's a chapter in the much larger story about how automation is actually reshaping work in real time. The narrative of robots replacing workers is seductive and simple. The reality is messier, more complicated, and ultimately more interesting.
Waymo has built extraordinary vehicles that can navigate complex urban environments, interpret traffic signals, anticipate pedestrian behavior, and execute thousands of real-time driving decisions every second. These vehicles are genuinely remarkable from a technical perspective. They represent decades of machine learning research, computer vision breakthroughs, and engineering excellence.
Yet they can't handle a passenger leaving a door partially open. This isn't a failure of technology. It's a revelation about the difference between automation and autonomy. The vehicles can automate driving. They can't autonomously handle all situations that arise in the real world.
Rather than pretending the problem doesn't exist or investing billions in solving it, Waymo opted for pragmatism. Acknowledge the limitation. Distribute the exception handling. Continue improving the core technology. It's honest systems design.
The door-closing pilot also reveals something important about economics and labor in the age of automation. Gig platforms like Door Dash have created infrastructure that's economically useful far beyond their original purpose. They've built payment systems, worker networks, geographic intelligence, and algorithmic task distribution systems. The marginal cost of adding a new task type (like door-closing) to an existing platform is near zero.
This creates economic incentives for companies to distribute more work across gig platforms rather than employing dedicated staff. It's efficient from a capital perspective. But it also fragments work further, increasing insecurity and reducing benefits. It shifts risks from companies to workers.
For workers, the door-closing task is neither a catastrophe nor a solution. It's additional income, available optionally, without long-term commitment. Some drivers will welcome it. Others will find it annoying. Most will be neutral, grabbing the occasional task when convenient.
The bigger question is what the pilot represents. Is it a temporary solution to a specific problem? Or is it the beginning of a broader pattern where gig workers become the exception handlers for automated systems across multiple industries?
The evidence suggests it's the latter. As companies deploy more automation, they discover more edge cases that automation can't handle. Rather than solving each edge case with dedicated technical solutions (which is expensive and slow), they'll distribute the exception handling across existing gig platforms (which is cheap and scalable). Workers will increasingly find themselves assembling income from multiple exception-handling microtasks distributed across multiple platforms.
This future isn't necessarily dystopian. It offers flexibility and multiple income sources. But it's also precarious, fragmented, and built on asymmetric power dynamics between platforms and workers.
Waymo's door-closing pilot in Atlanta is small. It probably affects a few hundred drivers and a few thousand passengers. But it's a microcosm of the future of work, automation, and human integration into economic systems being built right now. It's worth understanding not because it's shocking or novel, but because it's practical, realistic, and probably indicative of where the economy is heading.
The self-driving cars are mostly autonomous. The work isn't.
Key Takeaways
- Waymo pays DoorDashers $11.25 per task to close vehicle doors left ajar in Atlanta, revealing automation requires human exception handling
- Economic arbitrage at scale: distributing work across gig platforms costs 2.25M for dedicated employees
- Gig economy is fragmenting further: workers now assemble income from multiple microtasks across different platforms and companies
- Autonomous vehicles are highly automated (85-95%) for routine driving but still require humans for edge cases and exceptions
- Future economy will integrate AI and humans, with machines handling high-volume routine work and distributed gig workers handling exceptions
