Experts Say Self-Driving Cars Are the Elephant in the Room; We Still Have Time t

What “Problem” Does the “Solution” of Self-Driving Cars Actually Address?

Self-driving cars are presumed to be a panacea for traffic congestion, accidents, and pollution, but this premise itself needs serious scrutiny. The future envisioned by technological optimists often overlooks the complexity of transportation systems and their social costs. Do we really need another type of private vehicle to “optimize” already overburdened roads? Or is the real problem the urban structure overly reliant on private transport? The promotion of self-driving cars must first answer a fundamental question: Is it meant to perpetuate a “car-centric” development model, or serve as a transitional bridge toward a “human-centric” multimodal transportation system?

The current development path of the industry implicitly carries the risk of the former. Massive capital is pouring into autonomous perception and decision-making algorithms, but investment and attention on how to seamlessly and fairly integrate these vehicles into existing public transit networks are far from sufficient. This leads to an absurd status quo: we have AI that can recognize hundreds of objects, yet we cannot enable self-driving cars to achieve efficient, predictable interaction with a bus or a cyclist at a complex intersection.

The Fatal Gap Between Technological Maturity and Societal Acceptance

According to the latest annual report from the California Department of Motor Vehicles (DMV), the “disengagement rate” (the frequency at which a safety driver needs to take over) for major self-driving car companies in 2025 has improved by over 90% compared to five years ago, indicating significant technological progress. However, the annual survey conducted by the American Automobile Association (AAA) during the same period shows that the proportion of the public feeling “comfortable” riding in a fully self-driving car has only slowly increased from 15% to 28%.

This gap illustrates a key point: The bottleneck for the adoption of self-driving cars is shifting from hardware and algorithms to socio-psychological and governance aspects. A single accident amplified by social media can damage brand trust far more than millions of miles of accident-free data can endorse. The Silicon Valley model of “develop first, communicate later” has hit its ceiling in areas involving public safety and urban space.

mindmap
  root(Key Barriers to Societal Acceptance of Self-Driving Cars)
    (Media and Public Perception)
      (Amplification Effect of Single Accidents)
      ("Uncanny Valley" Psychology:<br>Nearly human but errors are less forgivable)
      (Distrust in "Black Box" AI Decision-Making)
    (Regulation and Liability Attribution)
      (Fragmented Cross-State/National Regulations)
      (Complex Accident Liability Chain:<br>Manufacturer, software provider, owner, passenger?)
      (Data Privacy and Ownership Disputes)
    (Infrastructure and Integration)
      (Inconsistent Road Signs and Quality)
      (Challenges of Mixed Traffic with Non-Autonomous Vehicles and Pedestrians)
      (Lack of Unified V2X Communication Standards)
    (Economic and Employment Impact)
      (Anxiety Over Disappearance of Professional Driving Jobs)
      (Impact on Related Industries from Changes in Vehicle Ownership Models)
      (Sharp Decline in Urban Parking Revenue)

Who Is Paying for the Future of Self-Driving Cars? Hidden Costs and Distorted Subsidies

The R&D and testing of self-driving cars is a capital-intensive gamble, and the source of funding and ultimate cost-shifting for this gamble will profoundly impact future transportation equity. Tech giants and automakers have invested hundreds of billions of dollars, and these investments inevitably expect returns. However, when we examine the business models of self-driving cars, we find their profit prospects highly depend on scaling, which may conflict with public interest.

First, there is the “free” use of public roads. Self-driving car test fleets extensively occupy roads for data collection and algorithm training, essentially turning public space into private R&D laboratories without paying corresponding “resource usage fees” or providing adequate compensation for the congestion and risks they cause. Second, to enable self-driving cars to operate smoothly, many cities are considering or have already invested in upgrading road infrastructure, such as clearer and more uniform lane markings, high-precision maps, and vehicle-to-everything (V2X) communication equipment. These expenses, often amounting to billions, are borne by public taxes, but the primary beneficiaries are a few tech companies and their users.

Even more alarming is the “efficiency trap.” Proponents of self-driving cars often argue they can increase road capacity and reduce congestion. But the law of “induced demand” in transportation tells us that when travel becomes easier and cheaper (e.g., saving driving effort), it attracts more people to choose that mode, eventually filling the added capacity and even increasing total vehicle miles traveled (VMT). A simulation study conducted by the International Transport Forum (ITF) indicates that in scenarios with insufficient sharing, self-driving cars could increase vehicle miles in metropolitan areas by 15% to 20%, negatively impacting energy consumption and carbon emissions.

Are Self-Driving Cars the Enemy of Public Transit, or the Final Piece of the Puzzle?

This is the most critical strategic choice for urban planners today. Positioning self-driving cars as a replacement for private car ownership or as an extension of mass transit will lead to vastly different futures. Unfortunately, existing business models and consumer habits are pushing the industry toward the former. Services like Waymo and Cruise’s Robotaxi are essentially more efficient taxis; they compete with private cars and may erode the ridership of less profitable bus routes.

However, the real potential lies in the latter. Imagine a scenario: self-driving shuttles serve as “first-mile/last-mile” connectors, efficiently transporting community residents to subway stations or bus trunk lines; or providing flexible shared ride services during off-peak hours or low-demand periods, complementing fixed schedules. This requires deep data integration, fare integration, and service design collaboration between self-driving car operators and public transit agencies, rather than operating in silos.

flowchart TD
    A[Core Urban Transportation Goals:<br>Accessibility, Equity, Sustainability, Safety]
    
    B{Strategic Positioning Choice for Self-Driving Cars}
    
    B --> C[Position as "Private Car Replacement"]
    B --> D[Position as "Mass Transit Extension Service"]
    
    C --> C1[Business Model:<br>Personal subscription, premium Robotaxi]
    C1 --> C2[Potential Consequences:<br>Reinforces vehicle ownership mentality<br>and road competition]
    C2 --> C3[Impact on Urban Goals:<br>May harm equity and sustainability]
    
    D --> D1[Business Model:<br>Government/agency service procurement,<br>integrated fares and trip planning]
    D1 --> D2[Potential Consequences:<br>Enhances convenience and coverage<br>of mass transit usage]
    D2 --> D3[Impact on Urban Goals:<br>Positively promotes all goals]
    
    A -.->|Should guide toward| D

Metropolitan areas in Taiwan, such as Taipei and Kaohsiung, face challenges with inefficient bus route operations in remote communities. This is precisely where self-driving medium and small buses can add value. Instead of having self-driving cars compete with taxis for passengers in areas like Xinyi District, policy guidance and subsidies could encourage operators to provide fixed shuttle services connecting to subway terminal stations during weekday evenings or holidays. This would not only address real transportation pain points but also allow the technology to mature iteratively in relatively controlled environments, building public trust.

The Regulatory Race: How to Find a Dynamic Balance Between Innovation and Safety?

Global regulatory frameworks are currently in a “catch-up” state, forming a fragmented patchwork. Different regulations for self-driving car deployment, data localization requirements, and liability principles have been proposed by various U.S. states, EU countries, and Asian economies. This inconsistency, while giving companies room to choose the most lenient environments for testing (“regulatory arbitrage”), will ultimately hinder the unification of technical standards and the expansion of cross-regional services, harming industry development and consumer rights in the long run.

Future regulatory thinking must shift from “approving specific technologies” to “managing systemic risks.” This means:

1. Performance-Based Regulation: Not limiting technical approaches, but setting clear safety, privacy, and equity performance thresholds (e.g., accident rates in specific scenarios must be a certain percentage lower than human drivers).
2. Transparency Requirements: Mandating disclosure of key safety data (e.g., disengagement event contexts, system limitations) and audit trails of algorithmic decision logic, without revealing core intellectual property.
3. Dynamic Compliance: Establishing a concept similar to aviation’s continuous airworthiness, requiring self-driving systems to continuously comply with the latest safety standards via OTA updates throughout their lifecycle.
4. Establishing Sandbox Mechanisms: Allowing companies to test innovative business models and technologies under regulatory supervision within designated physical or virtual areas, accelerating learning and rule-making.

When developing its self-driving car industry, Taiwan should avoid an “all-or-nothing” mindset. It could reference approaches like the UK’s, actively promoting authorization for “specific roads and scenarios,” such as initially opening closed campuses, port areas, or specific road sections during nighttime low-traffic periods. This allows the industry to gather data and validate technology in real environments while keeping risks manageable. This is more pragmatic than outright bans or blanket permissions.

Conclusion: Leading the Elephant Out of the Room Begins with Redefining the Problem

Self-driving cars are undoubtedly one of the most transformative technologies of this century, but their success is far more than just a technical achievement. What we need is not blind worship of “autonomous driving,” but a rational pursuit of “better mobility.” Industry, government, and civil society must collectively guide this “elephant in the room” into the sunlight for an honest dialogue: What kind of future cities do we want? What role should self-driving cars play in them? Who bears the costs? Who enjoys the benefits?

The next five years will be a critical formative period. Instead of asking “When can self-driving cars replace human drivers?”, we should ask: “How can we use automation, connectivity, and sharing technologies to build a safer, more equitable, and more sustainable transportation system?” The answer may not be driverless cars filling the streets, but a smart ecosystem deeply integrating flexible self-driving services, efficient mass transit, and safe walking and cycling environments. Technology should be a tool serving this vision, not the vision itself. Starting this conversation now, there is still time, but the window is narrowing.

FAQ

Why do experts call self-driving cars the “elephant in the room”? This metaphor refers to self-driving cars being a huge, obvious yet deliberately ignored key issue in the industry. Focus is on technological breakthroughs while avoiding core contradictions like their vaguely defined fundamental problem, social costs, and whether they truly match real transportation needs.

What is the biggest challenge self-driving cars currently face? The biggest challenge is not purely technical, but “societal integration.” This includes fragmented regulatory frameworks, imbalanced infrastructure investment, insufficient public trust, and failure to demonstrate priority over mass transit and micromobility.

How will self-driving cars affect urban planning and real estate? Without proper planning, self-driving cars could exacerbate urban sprawl and competition for road space. Conversely, if integrated with shared and demand-responsive services, they could free up parking spaces for conversion into green spaces or housing, reshaping urban landscapes.

What strategy should Taiwan adopt in the wave of self-driving cars? Taiwan should leverage its strengths in ICT and precision manufacturing, focusing on “vehicle-road-cloud” system integration, specific-scenario commercial services (e.g., ports, campuses), and establishing a human-centric evaluation framework to avoid blindly following technological hype.

When can consumers truly experience safe, fully self-driving services? Achieving scaled, fully self-driving services without safety drivers in major metropolitan areas optimistically still requires 5 to 10 years. Key factors include handling “edge cases,” unifying cross-brand protocols, and clarifying insurance and liability attribution.

Further Reading

1. California DMV Autonomous Vehicle Disengagement Reports - https://www.dmv.ca.gov/portal/vehicle-industry-services/autonomous-vehicles/disengagement-reports/
2. AAA Annual Automated Vehicle Survey - https://newsroom.aaa.com/tag/automated-vehicle-survey/
3. International Transport Forum (ITF) Study on the Impact of Autonomous Vehicles on City Traffic - https://www.itf-oecd.org/impact-autonomous-vehicles-city-traffic

{
  "image_prompt": "A conceptual, futuristic illustration in a dark blue and cyan color scheme. The"
}