Together with, Darian Thiemann, Darian Gorba, and Verena Kalwa I recently submitted our paper “Determinants of Acceptance of AI-Based Interaction Systems in Organizations – Two Vignette Experiments” to Computers in Human Behavior Reports.
In this paper, we examined data from the vignette experiments of Darian and Verena in depth.
Highlights
- AI systems become increasingly common as agents within organizations
- Perceived humanness of AI systems positively influences their acceptance
- How we think about AI as tools or social agents explain their acceptanc
- People want to have a say and favor participatory implementation of AI systems
- Fully autonomous AI decisions for personally relevant topics are not preferred
Abstract
For successful integration of Artificial Intelligence (AI) in organizations during technological progress, it is crucial to understand which psychological factors influence the acceptance of AI-based interaction systems. To contribute to this, we conducted two vignette experiments. In our first experiment (N = 196), we explored how different implementation strategies for assistive AI in organizations affect acceptance indicators such as warmth, competence, affective attitudes (fear and trust), and perceived humanness. Consistent with our hypotheses we found that participative and transparent implementation strategies lead to significantly higher acceptance compared to top-down approaches. Further, according to our hypotheses informed by Self-Categorization Theory and the Stereotype Content Model, we found indirect effects of warmth and competence when predicting trust and acceptance by humanness. In our second experiment (N = 288), we investigated applicants’ perceptions of AI in personnel selection by varying the degree of AI involvement in application decision. Influenced by factors such as humanness and social presence, the intention to use as a measure of acceptance decreased with increasing involvement. Additionally, following Unified Theory of Acceptance and Use of Technology, we found indirect effects for performance expectancy and social influence between humanness and behavioral intention. Our research highlights that both, the method of implementation and the extent of AI involvement, impact user acceptance levels. These findings contribute to the ongoing conversation about technology acceptance, particularly concerning AI applications in professional settings and provide indications for strategic approaches for its implementation in organizations.
Preprint
Part of the submission process was to publish the manuscript as a preprint on SSRN.
đź’ľ View abstract and download the preprint here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4911256