DAVI on Ethics

On Tuesday March 15, DAVI organized a colloquium with 25 professionals about the ethics of automated driving systems. The group talked about the responsibility of actions made by self-driving vehicles. Bart van Arem, professor at the Technical University of Delft, presented two different sides of a self-driving vehicle. On the one hand he showed the cuteness of a self-driving vehicle, the comfort and luxury. And on the other end he sketched an image of ‘a black killer car’ which was programmed to make decisions on life and death.

The second presentation was by Tom Alkim of Rijkswaterstaat. He talked about the developments in regulations of multiple countries in the EU to allow self-driving vehicles on the road. Safety is one of the biggest issues to implement this new way of transportation on existing roads. Research shows that self-driving vehicles causes up to 95% less accidents. But there will be all new kinds of accidents which are not caused by humans and that is hard to accept in our society.

Herman Wagter of Connekt presented the topic of ‘naughty software’. How can self-driving cars be integrated smoothly within normal traffic? Since both will be sharing the same roads, how naughty should we make our software for this transition period? Should we be teaching the machine right and wrong? For example, when a vehicle is parked half on the road and a self-driving car wants to pass, the software would not allow to cross the roadway in the opposite direction to go around the parking vehicle. In that case, the self-driving vehicle would cause a traffic jam, which is completely unnecessary. ‘Naughty software’ could allow the vehicle to adjust in mixed traffic and sometimes ‘break the rules’.

The final presentation came from philosopher Filippo Santoni de Sio of the Technical University of Delft. He enlightened the participants with his vision on ethics of automated driving systems. He mentioned the ethical questions in other branches in the past like military ethics. Those discussions are valuable to learn from. Filippo noticed that ethics of automated driving systems cannot be written by one player. Only by cooperating with multiple stakeholders, acceptance of mixed traffic with automated driving systems can be reached.

After these presentations the group was split into two groups. One on the ethics of automated driving systems and the other on ‘naughty software’. Interesting outcomes were recognition of self-driving cars in mixed traffic and a desire of balance between privacy and interpretation, safety and efficiency and safety and economic.