Introduction: Artificial intelligence (AI)-based text analysis offers new opportunities for assessing integrity, a critical yet elusive construct in human behaviours and interactions. Traditional measures such as personality inventories and cognitive tests provide valuable insights but often bias and miss the interplay between personality traits, moral judgment, and narrative expression. The accuracy of AI in this domain remains uncertain, raising the need for direct comparison with established psychometric tools.
Purpose: This study evaluates the accuracy and complementary value of AI text analysis in assessing integrity, benchmarked against the OMNI Personality Inventory and the Multidimensional Aptitude Battery (MAB).
Method: Data were collected from 10,218 Indonesian job applicants (aged 18+), who completed the OMNI, MAB, and a storytelling task. OMNI assessed 25 traits and 10 personality disorders, MAB measured cognitive and moral reasoning, and AI-driven natural language processing extracted integrity markers such as coherence, moral framing, and sincerity from narratives.
Results: Three integrity-related personality dimensions emerged from OMNI: Task Completion (Ambition, Dutifulness, Orderliness, Obsessive-Compulsive), Emotional Regulation (Antisocial, Impulsiveness), and Social Skill (Sincerity, Sociability, Warmth). MAB analyses revealed that applicants with psychopathy tendencies—identified by discrepancies between verbal and performance scores—performed well on general intelligence but scored significantly lower on integrity-relevant subtests (Comprehension, Picture Arrangement, Object Assembly). AI text analysis validated and extended these findings: applicants high in Task Completion and Social Skill produced coherent, prosocial narratives, while those low in Emotional Regulation or with psychopathy tendencies showed fragmented stories, weak moral justification, or self-contradictions. Importantly, AI detected integrity risks in some cases where OMNI and MAB results appeared within normal ranges, indicating added diagnostic value.


Conclusions: Integrity is best understood as a multidimensional construct integrating personality, cognition, and narrative moral reasoning. AI-based text analysis complements traditional psychometrics by uncovering subtle linguistic cues of integrity that standardized tests may overlook. Further validation is needed to refine accuracy and interpretability.