Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lee, Eldred, Worley, Nicholas, Takatsuji, Koshu
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2511.00774
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911248916938752
author Lee, Eldred
Worley, Nicholas
Takatsuji, Koshu
author_facet Lee, Eldred
Worley, Nicholas
Takatsuji, Koshu
contents This paper presents a retrospective analysis of anonymized candidate-evaluation data collected during pilot hiring campaigns conducted through AlteraSF, an AI-native resume-verification platform. The system evaluates resume claims, generates context-sensitive verification questions, and measures performance along quantitative axes of factual validity and job fit, complemented by qualitative integrity detection. Across six job families and 1,700 applications, the platform achieved a 90-95% reduction in screening time and detected measurable linguistic patterns consistent with AI-assisted or copied responses. The analysis demonstrates that candidate truthfulness can be assessed not only through factual accuracy but also through patterns of linguistic authenticity. The results suggest that a multi-dimensional verification framework can improve both hiring efficiency and trust in AI-mediated evaluation systems.
format Preprint
id arxiv_https___arxiv_org_abs_2511_00774
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Quantifying truth and authenticity in AI-assisted candidate evaluation: A multi-domain pilot analysis
Lee, Eldred
Worley, Nicholas
Takatsuji, Koshu
Human-Computer Interaction
Artificial Intelligence
This paper presents a retrospective analysis of anonymized candidate-evaluation data collected during pilot hiring campaigns conducted through AlteraSF, an AI-native resume-verification platform. The system evaluates resume claims, generates context-sensitive verification questions, and measures performance along quantitative axes of factual validity and job fit, complemented by qualitative integrity detection. Across six job families and 1,700 applications, the platform achieved a 90-95% reduction in screening time and detected measurable linguistic patterns consistent with AI-assisted or copied responses. The analysis demonstrates that candidate truthfulness can be assessed not only through factual accuracy but also through patterns of linguistic authenticity. The results suggest that a multi-dimensional verification framework can improve both hiring efficiency and trust in AI-mediated evaluation systems.
title Quantifying truth and authenticity in AI-assisted candidate evaluation: A multi-domain pilot analysis
topic Human-Computer Interaction
Artificial Intelligence
url https://arxiv.org/abs/2511.00774