Engineering · Human factors
The Car That Made You Worse at Driving
Modern cars prevent crashes brilliantly. They also ask a less practised, more distracted human to rescue them at precisely the moment the brilliance runs out.
Fewer pixels. Less work.
If you have ever wanted to feel less competent at driving, buy a more intelligent car.
In a 2005 Volvo V70, switching on the heated seat, lowering the temperature, changing the radio station and resetting the trip computer is not an exercise in information architecture. Your hand goes to a place. Your fingers find a shape. The road remains the largest object in your field of vision.
Put the same four tasks inside a modern display and they acquire menus, states, animations and the possibility that yesterday’s software update has moved something. The car has not merely replaced buttons with glass. It has converted a physical habit into a visual search.
Then, having taken your attention, it offers lane assistance to correct the result.
The machine is safer.
The human is not necessarily better.
Both can be true.01 · The interface
The screen did not remove the button. It removed touch.
In 2022, Swedish magazine Vi Bilägare asked drivers to learn the controls of twelve cars, then perform the same four everyday tasks while travelling at 110 km/h. The old Volvo completed them in ten seconds. The slowest modern car required 44.9.
That is not a universal laboratory verdict on every touchscreen. It was one controlled magazine test, and some screen-equipped cars did relatively well. Its useful finding is more specific: interface design matters enormously, and a newer interface is not automatically a faster or safer one.
Four identical tasks · 110 km/h · after familiarisation
NHTSA’s familiar warning says that reading or sending a text can take the eyes off the road for five seconds. At 55 mph, the car covers roughly the length of an American football field. The touchscreen test was not texting and the driver was not necessarily looking away for every second, so multiplying that slogan into “1.3 kilometres blind” would be dishonest.
But the underlying problem survives the correction. A control that demands repeated glances, confirmation and menu navigation consumes visual attention that a shaped switch does not.
44.9seconds to perform four routine tasks in the slowest car tested.
The correction from Brussels
Euro NCAP did not ban touchscreens.
It did something more measured. Its 2026 assessment gives human– machine interaction a formal place inside the new Safe Driving score. The protocol looks at the location, identification and ease of use of essential controls. Several functions — including the horn, indicators, wipers, high beam and hazard lights — are expected to have direct physical operation; other functions may still use direct touch or tightly limited menu paths.
A car does not simply “lose one star” because its fan speed lives on a screen. Scores and minimum thresholds across four safety stages determine the overall rating. But the direction is clear: crash safety now includes the interface used before the crash.
02 · The assistance
Now for the inconvenient fact: it works.
Lane-departure warning is not decorative technology. An IIHS study of police-reported crashes across 25 US states associated it with an 11 per cent reduction in relevant crashes and a 21 per cent reduction in relevant injury crashes after controlling for driver demographics.
Automatic emergency braking, blind-spot detection and other crash-avoidance systems have their own measured benefits. A serious argument cannot pretend that the old analogue car was safer merely because its controls were satisfying.
The modern car sees hazards you miss, brakes before you do and notices the lane marking during the instant in which the modern car’s menu has persuaded you to look elsewhere.
03 · The role
Driving has become supervision with occasional violence.
Level 2 assistance occupies an awkward legal and psychological category. The car can control speed and steering, yet the human must continuously monitor the road and remain ready to intervene. The driver is relieved of much of the repetitive work without being relieved of responsibility.
This creates what automation researchers call an out-of-the-loop problem. Active control supplies a continuous stream of small information: steering pressure, closure rate, road position, what the tyres are doing and what the driver is about to do next. Supervision is quieter. When the system reaches its limit, it asks the human to rebuild that situation quickly.
The evidence is not simple enough to claim that lane assist literally erases driving skill after fifteen minutes. Some research finds reduced workload and lower arousal; other longitudinal work suggests certain assistance can keep more attention on the forward roadway. Driver understanding can also improve with useful exposure to system limits.
What is well established is the mismatch: partial automation can encourage overreliance while still requiring immediate human competence. In one AAA-backed naturalistic study, drivers became more comfortable over six to eight weeks and received more attention warnings as time passed — consistent with more non-driving activity once familiarity grew.
04 · The paradox
The system can improve while the relationship deteriorates.
Crash avoidance
The car catches errors, watches blind spots and intervenes faster than a surprised human.
Interface load
Basic controls increasingly demand vision, precision and navigation through changing software.
Role confusion
Assistance looks like autonomy while the handbook quietly assigns every remaining failure to the driver.
Rare practice
The human performs less continuous control but remains the emergency fallback for situations the system cannot solve.
The point
Do not make the car stupider.
Keep automatic emergency braking. Keep blind-spot warning. Keep the systems that prevent a tired driver from drifting across a line. A machine that saves a collision does not become bad because its owner misunderstands it.
Make the division of labour honest instead. Give frequent controls a fixed location and tactile identity. Monitor driver attention in a way that cannot be defeated by resting a hand on the wheel. Describe assistance as assistance, not as a cheaper preview of autonomy. Teach owners where the system fails, not only how beautifully it works on a clean motorway.
Most importantly, stop designing the human as a spare component that can sit idle indefinitely and then perform perfectly on demand.
The old car asked more of you every second. The new one asks less until, suddenly, it asks everything.
That does not make the old car safer. It makes the new relationship harder to design than the industry’s language admits.
We built cars that think faster than us.
We should make sure they do not train us to think slower.
Fact check
Sources and necessary caveats
The article distinguishes proven crash reductions from the more conditional human-factors evidence around attention, trust and control handover.
- 01Vi Bilägare: physical controls and touchscreen task test, including full results and method
- 02NHTSA: distracted driving and the five-second eyes-off-road comparison
- 03Euro NCAP: 2026 protocols and the new Safe Driving assessment
- 04Euro NCAP: Safe Driving Driver Engagement protocol, version 1.0
- 05IIHS: measured crash reductions associated with lane-departure warning
- 06IIHS/HLDI: limited real-world evidence that partial automation adds benefits beyond crash-avoidance features
- 07AAA Foundation: how driver attention changed over six to eight weeks of ADAS use
- 08AAA Foundation: driver arousal and workload under Level 2 partial automation
- 09FHWA: longitudinal research on driver adaptation to vehicle automation