Mitigating Conflict from Cultural Communication Differences
The cross-cultural communication project explores misunderstandings which occur in intercultural online communication. We explore how OpenAI models can understand and resolve these conflicts using systematic prompt engineering. The ultimate goal for this project is to build a system that actively reduces cross-cultural misunderstanding, making online communication across the world more inclusive.
Problem
Cross-cultural interactions are increasingly common with technological advancement. Although this provides opportunities for creativity and collaboration, conflict may also arise due to differences in culturally influenced communication styles.
For instance, a Northern European manager may expect direct feedback, while colleagues from Asian cultural backgrounds may express disagreement indirectly, leading to unspoken concerns and misunderstandings. To mitigate this, our team is looking into incorporating large language models (LLMs) tools for digital communication.
Current Research Efforts
- Gathering culturally different digital data to create scenarios that accurately describe conflict felt in digital communication, which will be used in later studies to test LLMs
- Testing OpenAI’s GPT versions to see variance in LLMs across different models.
Publications
- Weng, Tzu-Yu, et al. "Identifying Tensions that Arise in Cross-cultural Online Communication." IUI Workshops. 2024.
- Weng, Tzu-Yu, et al. "Strangers in a new culture see only what they know": Evaluating Effectiveness of GPT-4 Omni for Detecting Cross-Cultural Communication Norm Violations." Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization. 2025.