Human-AI Persona Interaction Dataset: Chat Logs, Psychophysiology, and Survey Data from a Controlled Leadership Style Experiment

Read full text
Authors: Leon Ciechanowski, Anna Kovbasiuk, Konrad Sowa, Tamilla Triantoro, Aleksandra Przegalinska, Richard Freeman
Cited by: 2

Abstract

The dataset contains materials and data from an experiment examining the impact of an AI chatbot persona (a supportive "Servant Leader" persona vs. an antagonistic "Dark Triad" persona) on users' psychological, physiological, and behavioral outcomes, as well as task performance in a marketing collaboration task.The dataset includes data from pretest and posttest questionnaires, expert ratings of task performance, task products, psychophysiological data (electrodermal activity and facial electromyography), and anonymized chatbot conversation logs. The dataset may also include derived variables related to behavioral coding (including LLM-assisted coding results) used in analyses.The study was conducted using a cross-group design. The purpose of the deposit is to ensure transparency, replicability, and the possibility of further analyses in the areas of human–AI interaction, AI persona, psychophysiology, and creative collaboration.The dataset is associated with the manuscript entitled "The System's Shadow: How AI Personas Drive Psychological and Physiological Outcomes."

Publication details

Journal / source
RepOD
Publication date
1 January 2026
Document type
Dataset