Self-driving cabs may function acceptably on the wide, sunlit grid roads of the American South, but even in this relatively simple environment, they frequently struggle.
Would you trust them to navigate Bealach na Bà in the midst of a snowy Scottish winter? If placing a single traffic cone in front of their sensors can leave them paralyzed, how well do you think they’d handle a typical Polish roadwork site?
As a truck driver, I have the dubious privilege of operating modern trucks equipped with the latest driver assistance systems. Yet, these systems are far from ready to handle the complexities of daily driving.
This sentiment is echoed by nearly every truck driver I’ve spoken to.
While these systems can indeed save lives such as preventing a fatigued driver from rear-ending another truck in a motorway traffic jam they fall apart when faced with more demanding conditions, like single-track roads in the Scottish Highlands or an average Finnish winter day.
Drivers who don’t just cruise the motorway all day are already well-acquainted with their truck slamming on the brakes for no apparent reason sometimes multiple times per day.
The newest Mercedes Actros I drove last year in Finland had a particularly dangerous quirk: it would suddenly accelerate on its own while rounding a bend. My suspicion is that its speed limit data was outdated.
Keep in mind, this comes from the same company that claimed to have designed a fully autonomous vehicle four decades ago!
Artificial Intelligence Will Solve the Problem?
Proponents of AI argue that these issues will soon be resolved provided companies can collect enough “real-world” data. “Just let these cars drive around, and they’ll learn quickly,” they say.
However, there’s a major issue with this approach: agreeing to it essentially means allowing tech companies to use all of us as unwilling test subjects, with little to no government oversight.
And people are already dying because of it. Worse still, most of these companies never even asked for our consent to be part of their experiments.
And they aren’t teaching their vehicles to obey the rules of the road. Tesla has or at least had an option that allowed users to set the autopilot to exceed speed limits by a specified percentage.
Meanwhile, reports have surfaced indicating that Waymo cabs do not yield to pedestrians even when legally required to do so.
Some suggest that this is by design, allowing them to remain competitive with human drivers, who also often fail to yield.
A company spokesperson denied this, stating instead that it’s an “opportunity for continued improvement in these highly dynamic social interactions between man and machine.”
While it’s great that machines can learn, if they injure or kill someone in the process, would that be any consolation to the victim or their grieving family?
Legal and Ethical Dilemmas
Tech companies claim that self-driving cars are involved in fewer crashes than human drivers, but independent research contradicts this assertion. Evidence suggests that some automakers underreport incidents or attempt to obscure them from statistical records.

Tesla, for example, reportedly deactivates its self-driving system just seconds before an unavoidable crash. As a result, the official logs indicate that the accident occurred while the vehicle was under human control.
This could be a tactic to sidestep legal liability. After all, the question of who is responsible when an autonomous vehicle violates traffic laws or causes a crash remains unresolved. Is it the vehicle’s owner, the manufacturer, or the software developer?
Then there’s the infamous trolley problem: how should a self-driving car react if forced to choose between hitting different groups of people?
Should it prioritize children over the elderly, or vice versa? Should it protect its paying passengers at the expense of pedestrians? And if not, would you willingly buy or ride in a vehicle programmed to sacrifice your life to save others?
Congestion and Self-Driving Cars
Let’s not ignore the bigger issue: despite what its advocates claim, self-driving technology will not resolve the congestion that plagues our cities. Traffic is traffic, whether your vehicle is a two-stroke Trabant or an autonomous electric pod.
In fact, I believe self-driving technology will make congestion even worse. The idea is that after dropping you off, the self-driving vehicle will continue to another pickup.
But unlike personal cars, which remain parked while their owners work, fleets of autonomous taxis and minibuses will remain in circulation, constantly searching for passengers.
To meet peak-hour demand, cities will need an oversupply of these vehicles. But for much of the day, many of them will have nothing to do.
The result? Either they’ll cluster in parking lots on the outskirts of town, or they’ll aimlessly circle the streets, adding unnecessary mileage, noise, and pollution from tires and brakes.
You may think I’m a Luddite, but don’t just take my word for it. Listen to Elon Musk, one of the biggest champions of self-driving technology.
At a recent Tesla CyberCab premiere, he proudly pointed out a stretch of road packed with 20 driverless vehicles each one completely empty. I rest my case.
Build a Tram
Self-driving vehicles may find a niche role. They could provide a viable on-demand transport solution in rural areas if AI ever becomes advanced enough to handle the challenges of winding, narrow country roads in all weather conditions.
But outside of that specific scenario, why are we investing in automating buses with cutting-edge yet unreliable technology?
We already have a much simpler and more effective way to keep public transport vehicles on a fixed route: two parallel metal rails embedded in the street.
Regardless of the problem we’re trying to solve be it driver shortages, accessibility, environmental concerns, safety, or efficiency tracked vehicles outperform most alternatives.
And automating them is significantly easier, as anyone familiar with London’s DLR or a model train set can confirm.
If you want to move large numbers of people quickly through a dense city, the answer is clear: just build a tram!