Citation

BibTex format

@inproceedings{Sadek:2026:10.1145/3816046.3816265,
author = {Sadek, M and Mougenot, C},
doi = {10.1145/3816046.3816265},
title = {Whose Values? Demographic Influences on Perceptions of Conversational AI Value Alignment},
url = {http://dx.doi.org/10.1145/3816046.3816265},
year = {2026}
}

RIS format (EndNote, RefMan)

TY  - CPAPER
AB - Efforts to align conversational AI systems with human values often overlook whose values are being prioritised. This paper examines how demographic factors shape perceptions of conversational AI value alignment. It does so through a secondary analysis of survey data from 269 participants across 34 countries, revealing a previously overlooked dimension of alignment evaluation. Participants rated conversational AI transcripts along dimensions of value alignment, and responses were analysed with ANCOVA and post-hoc tests. Results show that demographic variables such as nationality, country of residence, ethnicity, and age significantly influence perceptions of alignment. They also highlight how aggregation can hide demographic divergence. These findings (i) demonstrate that value alignment is not a universal property of conversational AI, but is contextually situated, and (ii) highlight the importance of more mindful and inclusive participant sampling during conversational AI user evaluations, especially in relation to users' values.
AU - Sadek,M
AU - Mougenot,C
DO - 10.1145/3816046.3816265
PY - 2026///
TI - Whose Values? Demographic Influences on Perceptions of Conversational AI Value Alignment
UR - http://dx.doi.org/10.1145/3816046.3816265
ER -