Identifying context-specific values supports the engineering of intelligent agents that can elicit human values and take value-aligned actions. The proposed hybrid (human and AI) methodology simplifies the abstract task of value identification as a guided value annotation process by human annotators supported by natural language processing.

This work proposes a method to estimate a person’s relative preferences to a given set of values in the context of the ultimatum game. Further, it proposed two methods to reduce ambiguity in profiling the value preferences of these agents via interaction.

Bridging socio-technical gaps by proactively engaging societal stakeholders to translate values embodied by human rights into context-dependent design choices through a structured, inclusive, and transparent process.

  • Van der Waa, J., Van Diggelen J., Siebert, L. C., Neerincx, M., Jonker, C. (2020) Allocation of Moral Decision-Making in Human-Agent Teams: A Pattern ApproachProceedings of the Interational Conference on Human-Computer Interaction 2020 (HCII 2020). In: Harris D., Li WC. (eds) Engineering Psychology and Cognitive Ergonomics. Cognition and Design. Lecture Notes in Computer Science, vol 12187. Springer, Cham. 

In the foreseeable future practice, agents must collaborate with humans to ensure that humans always remain in control, and thus responsible over morally sensitive decisions. This work utilizes the concept of team design patterns to allocate moral decisions as a team task (agent and human), depending on properties such as the moral sensitivity, available information and time criticality.

This work discusses Value-Sensitive Design as an approach to design autonomous shipping systems in a manner that the systems can still be responsive to human operator's intentions and values.