Norms, Rewards, and the Intentional Stance: Comparing Machine Learning Approaches to Ethical Training

Daniel Kasenberg and Thomas Arnold and Matthias Scheutz

We compare our norm inference approach with inverse reinforcement learning for the task of learning moral and social norms from demonstrators that either are governed by explicit temporal logic representations of norms, or are reward-maximizing. We argue that our approach can be thought of as analogous to Dennett's "intentional stance".

@inproceedings{kasenbergetal18aies,
  title={Norms, Rewards, and the Intentional Stance: Comparing Machine Learning Approaches to Ethical Training},
  author={Daniel Kasenberg and Thomas Arnold and Matthias Scheutz},
  year={2018},
  booktitle={Proceedings of the 1st AAAI/ACM Workshop on Artificial Intelligence, Ethics, and Society},
  url={https://hrilab.tufts.edu/publications/kasenbergetal18aies.pdf}
}