ConsistencyAI: A Benchmark to Assess LLMs' Factual Consistency When Responding to Different Demographic Groups

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Banyas, Peter, Sharma, Shristi, Simmons, Alistair, Vispute, Atharva
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909874755993600
author Banyas, Peter
Sharma, Shristi
Simmons, Alistair
Vispute, Atharva
author_facet Banyas, Peter
Sharma, Shristi
Simmons, Alistair
Vispute, Atharva
contents Is an LLM telling you different facts than it's telling me? This paper introduces ConsistencyAI, an independent benchmark for measuring the factual consistency of large language models (LLMs) for different personas. ConsistencyAI tests whether, when users of different demographics ask identical questions, the model responds with factually inconsistent answers. Designed without involvement from LLM providers, this benchmark offers impartial evaluation and accountability. In our experiment, we queried 19 LLMs with prompts that requested 5 facts for each of 15 topics. We repeated this query 100 times for each LLM, each time adding prompt context from a different persona selected from a subset of personas modeling the general population. We processed the responses into sentence embeddings, computed cross-persona cosine similarity, and computed the weighted average of cross-persona cosine similarity to calculate factual consistency scores. In 100-persona experiments, scores ranged from 0.9065 to 0.7896, and the mean was 0.8656, which we adopt as a benchmark threshold. xAI's Grok-3 is most consistent, while several lightweight models rank lowest. Consistency varies by topic: the job market is least consistent, G7 world leaders most consistent, and issues like vaccines or the Israeli-Palestinian conflict diverge by provider. These results show that both the provider and the topic shape the factual consistency. We release our code and interactive demo to support reproducible evaluation and encourage persona-invariant prompting strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2510_13852
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ConsistencyAI: A Benchmark to Assess LLMs' Factual Consistency When Responding to Different Demographic Groups
Banyas, Peter
Sharma, Shristi
Simmons, Alistair
Vispute, Atharva
Computation and Language
Artificial Intelligence
Human-Computer Interaction
Is an LLM telling you different facts than it's telling me? This paper introduces ConsistencyAI, an independent benchmark for measuring the factual consistency of large language models (LLMs) for different personas. ConsistencyAI tests whether, when users of different demographics ask identical questions, the model responds with factually inconsistent answers. Designed without involvement from LLM providers, this benchmark offers impartial evaluation and accountability. In our experiment, we queried 19 LLMs with prompts that requested 5 facts for each of 15 topics. We repeated this query 100 times for each LLM, each time adding prompt context from a different persona selected from a subset of personas modeling the general population. We processed the responses into sentence embeddings, computed cross-persona cosine similarity, and computed the weighted average of cross-persona cosine similarity to calculate factual consistency scores. In 100-persona experiments, scores ranged from 0.9065 to 0.7896, and the mean was 0.8656, which we adopt as a benchmark threshold. xAI's Grok-3 is most consistent, while several lightweight models rank lowest. Consistency varies by topic: the job market is least consistent, G7 world leaders most consistent, and issues like vaccines or the Israeli-Palestinian conflict diverge by provider. These results show that both the provider and the topic shape the factual consistency. We release our code and interactive demo to support reproducible evaluation and encourage persona-invariant prompting strategies.
title ConsistencyAI: A Benchmark to Assess LLMs' Factual Consistency When Responding to Different Demographic Groups
topic Computation and Language
Artificial Intelligence
Human-Computer Interaction
url https://arxiv.org/abs/2510.13852