Starting around 2016, the idea of autonomous vehicles (AVs) sparked immense excitement. The vision of electric, self-driving cars just around the corner promised a revolution in transportation, ushering in an era of readily available robot services. Many, including myself at the time, envisioned a future reminiscent of Star Trek, where computers replaced drivers, vehicles were electric, and the economics of the automotive industry underwent a seismic shift. The thought was that million-mile cars, driven 24/7 by taxi companies, would dominate, making private car ownership financially impractical and personal driving obsolete, perhaps even illegal. Why buy a $100,000 vehicle destined to sit idle 95% of the time when you could access affordable, on-demand Robo Taxi services?
However, after eight years immersed in the connected vehicle industry, working with governments on the infrastructure to support new mobility solutions, my perspective has drastically changed. I’ve come to realize that groundbreaking technology doesn’t automatically translate into a viable business model in the real world.
AVs Are Harder Than We Thought
It turns out that teaching computers to drive is an incredibly complex challenge. While they excel in predictable scenarios, real-world driving is a constant negotiation of varied and unpredictable situations. Computers struggle with decision-making in unfamiliar contexts, as humorously highlighted by recent incidents of traffic cones disabling autonomous vehicles in San Francisco. This might not be simply a matter of needing more development time. There’s a growing suspicion that our current computing paradigms might not be ideally suited for true Artificial Intelligence, and that relying on glorified logic trees for driving may never be truly effective. We might need something closer to the nuanced processing power of a human brain – a capability we haven’t yet achieved.
The Public Policy Roadblock
Even if AV companies manage to weather the current venture capital drought and continue investing billions to refine the technology, a crucial question remains: Can they secure enough public acceptance to achieve widespread scaling? While many states permit autonomous vehicle testing and even commercial operations, often with minimal regulatory hurdles, numerous fundamental questions remain unanswered.
While legal liability and ownership issues, often cited as major obstacles, seem solvable, the broader regulatory landscape presents a significant challenge. Driving is already heavily regulated, encompassing everything from safety standards and fuel economy to charging infrastructure and roadway taxation. It’s unclear how robo taxis will profitably navigate this complex web of regulations.
The Forgotten Infrastructure Factor
One of the most pressing issues in transportation today isn’t vehicle technology, but infrastructure. Building and maintaining roads is incredibly expensive. While technology can offer some solutions, these too come at a cost. Transportation agencies at all levels grapple with funding shortages, and public appetite for tax increases to address these issues is limited. The gas tax, historically a primary revenue source for roadways, hasn’t been adjusted for decades, and finding alternative funding mechanisms faces significant political and practical hurdles. Exploring the concept of “road usage charging” reveals the policy complexities in this area.
My decade of experience in the field has made it clear that transportation challenges extend far beyond vehicle automation. Simply automating the car won’t even begin to address many of the systemic issues.
The Suitability of AVs as Taxis
Even if we assume that overcoming the technological hurdles of AVs is possible, and that infrastructure constraints are less limiting than they appear, the fundamental question of whether AVs make good taxis remains open.
The Overlooked Safety Aspect: Personal Security
When we discuss AV safety, the focus is usually on accident prevention. However, personal safety, particularly within the vehicle itself, is a critical consideration often overlooked. During a college internship in Washington D.C., a classmate taking a taxi home late one night experienced a near-rape by the driver and had her purse stolen during her escape.
Ride-sharing apps like Uber and Lyft offer some improvements by tracking rides, but they are not a panacea. Conversations with Uber and Lyft drivers reveal a litany of passenger-related incidents:
- Threats against drivers and other passengers
- Passengers passing out or experiencing medical emergencies
- Vomiting in the vehicle
- Offers of sexual favors instead of payment
- Sexual harassment and assault of drivers and other passengers
- Indecent exposure
- Vehicle damage
- Sexual activity during rides
- Unaccompanied minors being dropped off
- Rides booked for intoxicated individuals who become uncooperative or file kidnapping charges
- Crimes committed during rides (e.g., a passenger exiting the vehicle briefly to commit a shooting).
This list, compiled from firsthand accounts, is undoubtedly incomplete. While Uber and Lyft offer some safety improvements over traditional taxis due to driver tracking, they still rely on a human driver to monitor the vehicle and intervene in such situations. Removing the driver in a robo taxi scenario raises serious questions about passenger safety and security.
The Harsh Economics of Mobility as a Service
Remember the scooter craze led by companies like Bird and Lime? Scooters were initially hailed as the future of urban mobility – fun, electric, and a convenient alternative to cars or public transport. Mobility as a service seemed destined for widespread adoption. However, the reality has been far less rosy. Some scooter companies have gone bankrupt, while others face stringent regulations. Even those that remain struggle to achieve profitability. Making money in the mobility-as-a-service sector is significantly more challenging than many initially anticipated, especially for those without deep industry experience.
A major cost driver for scooter companies is vehicle repair. When users don’t own the vehicles, they tend to treat them with less care, leading to increased vandalism and theft. Furthermore, the logistical challenges and costs of constantly relocating scooters to balance supply and demand are substantial, particularly given the often-unidirectional nature of user trips.
The Robo Taxi Rebalancing Problem
Ontario Airport, located inland from Los Angeles, is my preferred airport despite being a 45-minute drive away. While driving and parking is often inconvenient due to wait times and costs, relying on ride-sharing services like Lyft presents its own drawbacks. Taking a Lyft home from the airport, I often feel for the driver. They earn a decent fare for the ride, but then find themselves far from urban centers late at night, facing a long, unpaid drive back home.
This “deadheading” dynamic, the empty return trip, is a critical factor that undermines the financial viability of robo taxis. While the idea of autonomous electric vehicles operating 24/7 sounds appealing, it overlooks the significant cost of network rebalancing. Just like with scooters, robo taxi networks will require constant rebalancing to meet demand. While robo taxis can theoretically drive themselves to rebalancing points, this still incurs costs in terms of time and energy. When factoring in vandalism, theft, and the need for remote monitoring and intervention to address safety and security concerns, the overall cost of robo taxi services is likely to be much higher than initially projected. This is further compounded by the need to generate profits for the venture capital firms currently funding these ventures.
If my assessment is even partially accurate, the cost of riding in a robo taxi might not be significantly lower than owning a personal Tesla with advanced driver-assistance systems, which offers the added benefits of private vehicle ownership, safety, convenience, and personal care.
While I remain a technology enthusiast and believe in the transformative power of computing, it’s crucial to recognize that not every problem is fundamentally a technological one. Sometimes, the non-digital aspects – the human, social, economic, and infrastructural – are the real challenges that must be addressed.
