Azrul Aziz

Hypothesis: 95% less will be killed in traffic in year 2040

Approximately 37 000 people die in road crashes each year in US. Countries that have completed the transition to 100% autonomous vehicles on ordinary roads by 2040 will have less than 5% deaths in traffic compared to today’s figures. In US that would mean that less than ~2000 people would die in traffic by 2040. Is there any substance to these figures? Let’s briefly look at why the figures are reasonable and then let’s scratch the surface on what other impact on the society autonomous cars might have in the future.

The substance 

94% av all accidents are caused by human error according stats from 2015 by NHTSA. The causes of the remaining 6% is relatively evenly distributed between vehicle, environment and unknown reasons. 

After replacing humans with computers the human errors will be replaced by new types of errors; errors related to the driving software and errors related to the driving hardware. Driving software is the software needed to drive the car from one place to another in a safe and energy efficient way. This driving software will improve continuously, for example by all autonomous cars contributing with data from real traffic situations. The driver hardware will be things like cameras, radars, other sensors and communication devices. As any other high-tech products the hardware capability will be improved over time. Since there are both business- and political reasons for not releasing autonomous cars that are unsafe it’s reasonable to think that these kinds of errors soon will be relatively limited. One could also argue that the total number of accidents will be decreased as a result of improvements in both vehicle and road safety. And we’ll probably see wheels communicating and cooperating with roads for increased efficiency before being replaced by a disruptive solution further into the future. The cars of the future will be able to drive both faster and safer at the same time. For how long will we still need to wear seat belts?

In other words, when autonomous cars entirerly replaces human drivers there will no longer be any accident of the sort that describes 94% of today’s accidents, ie human error. Some new types of accidents will probably happen due to the driver software and hardware but it’s reasonable to think that these will have a relatively small impact compared to today’s number of accidents, considering that the AI in a car never gets tired, never is distracted and rarerly makes a bad decision.

Autonomous cars will change the infrastructure in cities in ways that Dean Kamen promised that the Segway would, but failed to. In part 2 we will look at potential short term impacts on the society.