Whose Preferences? Differences in Fairness Preferences and Their Impact on the Fairness of AI Utilizing Human Feedback

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lerner, Emilia Agis, Dorner, Florian E., Ash, Elliott, Goel, Naman
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910479236988928
author Lerner, Emilia Agis
Dorner, Florian E.
Ash, Elliott
Goel, Naman
author_facet Lerner, Emilia Agis
Dorner, Florian E.
Ash, Elliott
Goel, Naman
contents There is a growing body of work on learning from human feedback to align various aspects of machine learning systems with human values and preferences. We consider the setting of fairness in content moderation, in which human feedback is used to determine how two comments -- referencing different sensitive attribute groups -- should be treated in comparison to one another. With a novel dataset collected from Prolific and MTurk, we find significant gaps in fairness preferences depending on the race, age, political stance, educational level, and LGBTQ+ identity of annotators. We also demonstrate that demographics mentioned in text have a strong influence on how users perceive individual fairness in moderation. Further, we find that differences also exist in downstream classifiers trained to predict human preferences. Finally, we observe that an ensemble, giving equal weight to classifiers trained on annotations from different demographics, performs better for different demographic intersections; compared to a single classifier that gives equal weight to each annotation.
format Preprint
id arxiv_https___arxiv_org_abs_2406_05902
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Whose Preferences? Differences in Fairness Preferences and Their Impact on the Fairness of AI Utilizing Human Feedback
Lerner, Emilia Agis
Dorner, Florian E.
Ash, Elliott
Goel, Naman
Machine Learning
Artificial Intelligence
Computation and Language
Computers and Society
There is a growing body of work on learning from human feedback to align various aspects of machine learning systems with human values and preferences. We consider the setting of fairness in content moderation, in which human feedback is used to determine how two comments -- referencing different sensitive attribute groups -- should be treated in comparison to one another. With a novel dataset collected from Prolific and MTurk, we find significant gaps in fairness preferences depending on the race, age, political stance, educational level, and LGBTQ+ identity of annotators. We also demonstrate that demographics mentioned in text have a strong influence on how users perceive individual fairness in moderation. Further, we find that differences also exist in downstream classifiers trained to predict human preferences. Finally, we observe that an ensemble, giving equal weight to classifiers trained on annotations from different demographics, performs better for different demographic intersections; compared to a single classifier that gives equal weight to each annotation.
title Whose Preferences? Differences in Fairness Preferences and Their Impact on the Fairness of AI Utilizing Human Feedback
topic Machine Learning
Artificial Intelligence
Computation and Language
Computers and Society
url https://arxiv.org/abs/2406.05902