The increase in the use of Artificial Intelligence (AI) algorithms in applications impacting human users and actors has, as a direct consequence, the need for endowing these AI systems by ethical behaviors. Several approaches already exist, but they fail to allow adaptability when facing new situations. To tackle this problem, we propose a new approach using Reinforcement Learning. We also propose a generic multi-agents simulator, that we adapt to the case of intelligent management of energy distribution in Smart Grids, and that we use in order to evaluate our implementation of this proposed approach. We then present the results using multiple rewards that reflects ethical behaviors.
This is the thesis I redacted at the end of my Master in Artificial Intelligence. The subject was Evolutive learning of ethical behaviors, realized at the LIRIS lab, under the supervision of Professor Salima Hassas and Professor Olivier Boissier.
The documents (thesis and slides) are in French, but the English abstract is available.