In the Netherlands, an average of approximately 625 people die in traffic each year. In many cases, this happens because someone loses control over their vehicle or movement. Others are sometimes unintentionally drawn into these incidents. Over the past decades, much has been done to reduce major causes of accidents—such as improving vehicles and making roads safer. There is now little additional gain to be made there. What remains is driver behavior. If we can change that—or even remove it from the process—the number of traffic fatalities could drop significantly. Many people agree with this in principle, but are we really willing to do everything necessary to achieve it? It does not seem so.
Facts about traffic fatalities
Compared with other countries, the Netherlands performs relatively well. In neighboring countries, the number of fatalities is about 30% higher. Since the 1970s, the number of traffic deaths in the Netherlands has declined, despite increasing traffic density. Since 2010, however, this number has stabilized.
Cyclists and car drivers are most often the victims, with roughly 250 deaths per year each. Notably, cyclists often die as a result of collisions with cars. As a result, car drivers are the largest contributors to fatal accidents. Training, education, and penalties for drivers appear to have little additional effect. The most logical next step is therefore to remove the driver from the process—through autonomous driving (autopilot).
Facts about experiments with autonomous driving
Many initiatives focus on self-driving taxis (robotaxis). China has projects in at least 20 major cities. Conditions there are favorable: strong political support, a large number of electric vehicles, and relatively low costs. Major players include Baidu and WeRide, which are given ample room to gain experience.
In the United States, at least ten trials are underway, including those by Tesla, Waymo (Alphabet), and Zoox (Amazon).
Europe clearly lags behind. In Germany, Mercedes is allowed to test on one specific stretch of highway. In the Netherlands, there are only small-scale experiments with self-driving buses. Large-scale trials with autonomous cars are unlikely to be permitted before 2027.
Why do so many deaths still occur?
1. Selective reporting
2. Fear of making mistakes
3. Self-overestimation
4. Learned skepticism
Are these fears justified?
Until recently, these concerns were understandable—the technology was not yet good enough. But developments are moving extremely fast. Data-center computing power is increasing explosively, and systems are improving rapidly as a result. Consider ChatGPT: two years ago, most people had never heard of it; today, many use it daily. We see the same trend in autonomous driving. There is now abundant availability of:
- vast amounts of real-world driving data
- massive computing power in central systems
- artificial intelligence
- fast onboard computers in vehicles
As a result, fear of autopilot is no longer rational. What once was useful caution is now holding us back. Daniel Kahneman calls this a bias. Within Brain Based Safety, we refer to this as a “kronkel”—something that was once useful, but now impedes us.
What are we waiting for?
What is most needed now is scaling up real-world testing. We must build trust through experience. That requires daring to make mistakes and learning in practice. Policymakers need to be encouraged to move faster. The paradox is that caution now costs lives: every day we wait, victims fall who could otherwise have been prevented.
In short, we must also look at ourselves. We must reprogram the biases in our own automatic pilot—even if that feels uncomfortable.
Juni Daalmans
November 2025