Conflict Solving Through Chatbot-Guided Reflection

Abstract

Introduction

Conflicts are a natural part of human interaction, but individuals often lack tools to address them constructively. This study explores whether a chatbot-guided reflection can foster conflict resolution by enhancing users’ self-awareness, emotional regulation, and clarity of perspective. AmicAI is a generative AI-powered chatbot designed around Moore’s Circle of Conflict [1], guiding users through six potential conflict sources: values, relationships, structure, data, moods, and interests. By engaging with AmicAI, users reflect on their current conflict from multiple angles, promoting better self-understanding and more effective problem-solving.  The research investigates if the interaction with the chatbot is perceived as helpful, whether it shifts the user’s understanding of the conflict, and whether it influences their readiness to act differently in similar situations.

Methods

The study targets Slovenian-speaking students (N = 50). Participants complete a pre-study questionnaire and the Conflict Resolution Questionnaire (CRQ) [2] that assesses participants' conflict resolution skills, then interact with AmicAI while reflecting on a current conflict of their choice, and complete a post-study questionnaire and again the CRQ. AmicAI, powered by GPT-4o and built in the OpenAI Playground in the first part of this project, guides users through Moore’s six conflict sources without retaining conversation history. Pre- and post-study questionnaires assess chatbot perception, engagement, and emotional impact using Likert scales (1–5) and open-ended responses. Quantitative and qualitative data are analyzed, with open-ended answers categorized for emerging themes.

Expected Results

A pilot (n = 12) showed improvement in the “Clarify Perceptions” dimension of the CRQ, especially among participants with favorable views of chatbot interaction. While other effects were limited in the pilot, the full sample is expected to show broader gains in emotional clarity, perspective-taking, and readiness to act.

Limitations

Limitations include the chatbot’s lack of memory, which restricts conversation continuity, and the self-selecting nature of the sample. The study also relies on self-report data, which may be subject to bias. Participants choose and freely discuss their own conflict, adding variability that must be considered in data analysis. 

Discussion

The preliminary results suggest that AmicAI can support emotional clarity and perspective-taking in conflict situations. Even a single interaction showed potential benefits. Therefore, we’re aiming towards a study with larger samples to confirm these effects, with the goal of understanding how chatbot-guided reflection can be applied to more complex conflict scenarios.

References

[1] C. W. Moore, The mediation process: Practical strategies for resolving conflict, 4^th edition. New York: John Wiley & Sons, 2014.

[2] M. Henning, "Evaluation of the conflict resolution questionnaire," M. S. thesis, Auckland University of Technology, 2003.