Trust in the machine: how contextual factors and personality traits shape algorithm aversion and collaboration
This paper studies how contextual factors and personal variables influence the delegation of decisions to an algorithm. Using a multi-armed bandit task, we conducted an experiment with four treatments - baseline, explanation, payment, and automation - where participants repeatedly chose between maki...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
March 2025
|
| In: |
Computers in human behavior reports
Year: 2025, Volume: 17, Pages: 1-17 |
| ISSN: | 2451-9588 |
| DOI: | 10.1016/j.chbr.2024.100578 |
| Online Access: | Verlag, kostenfrei, Volltext: https://doi.org/10.1016/j.chbr.2024.100578 Verlag, kostenfrei, Volltext: https://www.sciencedirect.com/science/article/pii/S2451958824002112 
|
| Author Notes: | Vinícius Ferraz, Leon Houf, Thomas Pitz, Christiane Schwieren, Jörn Sickmann |
| Summary: | This paper studies how contextual factors and personal variables influence the delegation of decisions to an algorithm. Using a multi-armed bandit task, we conducted an experiment with four treatments - baseline, explanation, payment, and automation - where participants repeatedly chose between making decisions themselves or delegating to an algorithm under uncertainty. We evaluated the impact of Big Five personality traits, locus of control, generalized trust, and demographics alongside the treatment effects using statistical analyses and machine learning models, including Random Forest Classifiers for delegation behavior and Uplift Random Forests for causal effects. Results show that payment reduces delegation, whereas full automation increases it. Age, extraversion, neuroticism, generalized trust, and internal locus of control significantly and consistently influenced delegation decisions across both predictive and causal analyses. Additionally, female participants reacted more strongly to algorithm errors. Increased delegation rates improved algorithm accuracy. These findings provide new insights into the roles of contextual conditions, personal variables, and gender in shaping algorithm aversion and utilization, offering practical implications for designing user-centric AI systems. |
|---|---|
| Item Description: | Online verfügbar: 28. Dezember 2024, Artikelversion: 1. Januar 2025 Gesehen am 16.01.2025 |
| Physical Description: | Online Resource |
| ISSN: | 2451-9588 |
| DOI: | 10.1016/j.chbr.2024.100578 |