Marseille edit
The report of the ethics committee on the automated and connected driving returned in June was presented this week by Federal Minister of Transport and digital infrastructures Alexander Dobrindt. Chaired by Doctor Udo di Fabio, the former judge of the German Constitutional court and maintaining professor of Bonn University, it groups experts of various disciplines going of the philosophy, the law, the social sciences to the evaluation of the impact of technologies, the car industry and the development of software. His objective is "to develop the ethical orientations necessary for the automated and connected driving".
The committee notes technological developments in automobile subject. "The partial automation has already been a standard" on the subject. Soon, the autonomous vehicles will be a reality. Yet, this evolution comes along with an improvement of the safety to the point that we "cannot exclude that at the end of this development, there will be motorized vehicles which will be intrinsically safe, that is to say, they will never be involved in an accident". However, the Committee qualifies by asserting that it "will not be possible to prevent completely accidents (…) given the realities of the heterogeneous and disconnected road traffic". It remains insufficient to reject, in principle, the autonomous cars. Their advantages "in terms of greater mobility, improved safety and saving of time overtaking prima vista the inherent risks to these systems".
This perspective has to urge the government to think about these subjects. It is the role of the committee that to enlighten the German power on the ethical stakes in the autonomous driving. As such, it emitted twenty recommendations, what would be the world’s first in this area.
The problem of the slightest evil
The panel of experts deals with in particular the "situations of dilemma". Concretely, you have driven aboard your autonomous car which detects several children playing on the road. What does your behavior have to be? To avoid the children at the risk of killing you or bringing them inside to protect your life? It is the dilemma of the streetcar that leads to the choice between two problems. "Easy", you tell me. "Both problems, I choose the slightest". Except that it is difficult to envisage in advance all the possible scenarios in their complexity. Yet, it is exactly for what we ask to the engineers and to the programmers of autonomous cars.
To help them, the committee emits a recommendation based in general principles. The right to life is imperative in any other considerations, particularly the property if this one may be damaged. It is advisable to specify that it is about the right to human beings’ life which gets the upper hand over that of the animals. But as sensitive beings, their protection has to prevail over that of the property.
In case of damage susceptible to affect two people or two groups of people, the committee is less categorical. In case of emergency, the vehicle does not have to choose the one rather than the other one. There is no "sacrifice" or "compensation" to be made because it would lead the IT specialists to introduce discriminatory criteria into their programs.
As such, the individual must be considered as "sacrosanct". We shall notice that it is not of a big help for the programmers. Moreover, the committee has no solution in case several lives are threatened in an imminent way. "The only thing concerned is to save so many innocent people as possible "it’s satisfied with asserting. Is it a matter of the simple arithmetic? Does the use of the "innocent" adjective mean that the life of one or several "convict(s)" (it could be the group of children which has to make nothing on a road) must not be favored? The experts kept in touch as waiting for detailed studies are driven. However, they add farther a precision indicating that "those involved in risks of mobility do not have to sacrifice those who are not them".
Security vs freedom
The report also focuses on the constraints given by the autonomous cars on the driver’s freedom of action. In an undistant period, the car was considered as a sign of independence (the object of the passage in the adulthood) if it is not as a symbol of freedom: it is Barry Newman turning darker at the top speed to escape the police in Zero Limit, Steve McQueen offering a memorable track race in Ford Mustang in Bullitt or even the bloody epic of Bonnie & Clyde. It is not sure that this romantic image continues with the autonomous cars.
If the committee praises their profits in term of safety, they have to agree with the principle of freedom. "There is no ethical rule which places [the first] before [the second"], it strikes without ambiguity. Also, the driver is free to turn or not with the possibilities offered by the automated procedures. These can assist it by recommending him/her not to take the steering wheel if he/she is tired but keeps the final privilege to decide.
Big dated, security and private life
Besides, panel is interested in the problem of the masses of data (big data). They are those concerning drivers (average speed, driving duration, fatigue state …), the others and their vehicles and those concerning the time, the road … General principle: the user has full sovereignty in his personal data on which he decides to protect or reveal as he/she sees fit. Nevertheless, he is not sure that this power can practice so easily in practice because of the complexity of these systems.
That is why the State has to intervene upstream by assuring a protection. It can pass by the granting of certifications concerning the automated functions of driving. Besides, the research on the anonymisation’s technique has to be the object of investments. Procedures and technical solutions have to allow the users to be informed about the state and the use of these databases. It is necessary to make sure that all the stakeholders agree to their use what requires a legal approach.
A scattered responsibility
Finally, thorny problem of the responsibility was raised. The autonomy asks to rethink about the car in term of system. There isn’t a machine piloted by a driver but a system concerning several actors: the vehicle, the road (equipped with information sources sensors), all the elements of the communication infrastructure …It results from a responsibility share which it is advisable to determine in the event of an accident. If the manufacturers are responsible for the functional safety of the systems, telecoms operators can be pointed in case of failure in the data transmission. In this respect, we can envisage the creation of a catalog updated of scenarios on the basis of the listed errors. The idea would be that the vehicle can refer to this document and deactivate the automated diving functions in case the situation corresponds to a scenario already seen. The machine could also update constantly to avoid the risks of errors, following the example of any computer system.
After all, ecognition should be given to the valuable work of this communication of ethics which exposes clearly the difficulties provided by the autonomous vehicles and asks sometimes more questions which it does not answer. A report must be reversed as quickly as possible because the technological progress, does not wait.
Article written by Thierry Randretsa