-
Amanda MAHABIR
At a time when intelligent robots are increasingly efficient and sought after, one may wonder what civil liability regime would apply to artificial intelligence if an accident or failure were to occur.
Intelligent robots are now part of our daily lives and generate significant interest.
Robots assist surgeons, help them make diagnoses, are capable of managing home energy, anticipate our needs, and monitor our personal data.
Determining what the liability regime would be seems, in practice, very difficult to establish given the specific nature of the intelligent robot.
The robot is an object that has the capacity to make decisions via algorithms that can be compared to human neurons. Thus, the robot is capable of analyzing its surroundings and having a certain reasoning, allowing it to constantly improve. Indeed, algorithms allow the robot to learn by itself, to know how to act in certain situations, and to make choices.
In this way, the designer can hardly anticipate the behavior of an intelligent robot.
So who is responsible for the actions of a defective robot?
The European Parliament insists on the fact that technological deployments must be taken into account without neglecting the potential of artificial intelligence and its dangers.
Currently, the project to reform the law of civil liability does not grant any specific place to intelligent objects.
Therefore, we will look at pre-existing liability regimes to find out if one of these regimes is applicable to intelligent robots.
- Liability for things (Strict liability)
The custodian is responsible for the things they have under their guard, as they are supposed to know how to control them. There is therefore a link between the custodian and the act of the thing.
A robot in motion or not can cause damage.
A robot that exceeds the instructions given to it and acts autonomously can cause accidents. This autonomy must therefore be considered from the perspective of a defect in the thing or abnormal behavior of the robot.
The robot is thus a thing, and the act of the robot thing is under the responsibility of its custodian—the user of the robot and not the creator (custodian of behavior).
- Product liability
The robot can be qualified as a product equipped with artificial intelligence whose acts can cause damage. As a result, the robot can be considered a defective product that does not offer the safety guarantee that one can expect.
A defective robot causes damage due to a runaway of its intelligence.
Also, the manufacturer of a finished product or a component is responsible for the defectiveness of the robot.
Shortcomings of pre-existing regimes
The regimes of liability for things remain difficult to apply to intelligent robots, given their evolution and autonomy.
At the same time, the behavior of a robot does not seem to be in line with the notion of defectiveness in the legal sense as we understand it. Damage can be caused without the producer being responsible, provided that in the state of their scientific and technical knowledge at the time the product was put into circulation, they could not detect the existence of a defect.
As the robot is endowed with immense potential, the defectiveness originates in the progress of the machine itself, in its adaptability to its environment, and through its emancipation. Often, the creator is overtaken by the robot they created, because this machine can ultimately act alone.
Exonerating the manufacturer from all liability given the development risk of intelligent robots does not seem adapted to technological evolution.
Thus, positive law is not yet able to designate the person responsible for damage caused by artificial intelligence.
Therefore, it will be necessary to think of a new liability framework taking into account the damage caused by an intelligent robot, and by revaluing the development of artificial intelligence and the role played by those who market these technological advances.
It would be a liability regime for things applicable to intelligent objects highlighting safety standards.
It would also be necessary to determine which behaviors could cause damage and require manufacturers to provide coding lines to prevent potential dangers or integrate control by a human person.
In this way, the potential damage caused would engage the liability of the program designer if the damage occurred following negligence of the custodian of the structure.
Failing to be able to engage their liability, the damage should be covered by a compensation fund since these robots lead us back to the notion of acceptance of development risks by society.
The debate regarding the liability regime remains open and, to date, the place of risk in our societies raises many questions.
