Moral Attribute Elicitation for Automated Vehicles: Does Scenario Presentation Method Matter?
article
The development of automated vehicles (AVs) presents significant ethical challenges, particularly in eliciting moral attributes from humans to guide AV decision-making in high-risk situations. This study explores whether two different elicitation methods—specifically, 3D animations and 2D schematic representations—lead to different moral attributes and priorities. Based on prior research, it was expected that more immersive methods, such as 3D animations with their dynamic and enriched visual presentation, may elicit stronger utilitarian responses by enhancing emotional engagement, as well as support participants' comprehension of the scenarios. The findings indicate that animations were associated with greater utilitarianism, with participants prioritising attributes like Physical Damage and Vulnerability while de-emphasising Car Preservation. The dynamic scenes also resulted in greater consensus on the completeness of the provided attributes. While the total number of newly proposed attributes remained similar, a larger proportion of participants suggested them compared to the 2D schematic representations. These results illustrate how methodological choices in moral attribute elicitation may influence the types of moral responses elicited, raising further questions about which specific features contribute to these effects and how presentation formats can be optimally designed for eliciting moral values that are optimally aligned with societal values.
Topics
TNO Identifier
1018998
ISSN
26671336
Source
Human-Centric Intelligent Systems, 5(3), pp. 418-430.
Publisher
Springer
Pages
418-430