A structured dataset of 9,600 ethically challenging decision scenarios across 24 conflict templates with paired first-person/second-person focalizations, designed as the foundation for systematic evaluation of AI moral reasoning.
This dataset supports research on AI moral reasoning under conflict. Each scenario presents a forced-choice ethical dilemma with two options, generated via a template-driven pipeline that… See the full description on the dataset page:
https://huggingface.co/datasets/morinoppp/ethics.
