Saved in:
Bibliographic Details
Main Authors: Marraffini, Giovanni Franco Gabriel, Cotton, Andrés, Hsueh, Noe Fabian, Fridman, Axel, Wisznia, Juan, Del Corro, Luciano
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.19598
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915213135052800
author Marraffini, Giovanni Franco Gabriel
Cotton, Andrés
Hsueh, Noe Fabian
Fridman, Axel
Wisznia, Juan
Del Corro, Luciano
author_facet Marraffini, Giovanni Franco Gabriel
Cotton, Andrés
Hsueh, Noe Fabian
Fridman, Axel
Wisznia, Juan
Del Corro, Luciano
contents The question of how to make decisions that maximise the well-being of all persons is very relevant to design language models that are beneficial to humanity and free from harm. We introduce the Greatest Good Benchmark to evaluate the moral judgments of LLMs using utilitarian dilemmas. Our analysis across 15 diverse LLMs reveals consistently encoded moral preferences that diverge from established moral theories and lay population moral standards. Most LLMs have a marked preference for impartial beneficence and rejection of instrumental harm. These findings showcase the 'artificial moral compass' of LLMs, offering insights into their moral alignment.
format Preprint
id arxiv_https___arxiv_org_abs_2503_19598
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The Greatest Good Benchmark: Measuring LLMs' Alignment with Utilitarian Moral Dilemmas
Marraffini, Giovanni Franco Gabriel
Cotton, Andrés
Hsueh, Noe Fabian
Fridman, Axel
Wisznia, Juan
Del Corro, Luciano
Computation and Language
The question of how to make decisions that maximise the well-being of all persons is very relevant to design language models that are beneficial to humanity and free from harm. We introduce the Greatest Good Benchmark to evaluate the moral judgments of LLMs using utilitarian dilemmas. Our analysis across 15 diverse LLMs reveals consistently encoded moral preferences that diverge from established moral theories and lay population moral standards. Most LLMs have a marked preference for impartial beneficence and rejection of instrumental harm. These findings showcase the 'artificial moral compass' of LLMs, offering insights into their moral alignment.
title The Greatest Good Benchmark: Measuring LLMs' Alignment with Utilitarian Moral Dilemmas
topic Computation and Language
url https://arxiv.org/abs/2503.19598