Skip to main navigation Skip to search Skip to main content

Generative AI and the Future of Cross-cultural Dialogue?

Research output: Contribution to conference › Conference abstract › peer-review

Abstract

What are the mechanisms by which generative AI based on large language models engenders bias and what might these signal for the future of cross-cultural dialogue? This presentation focuses on the effects which the adoption of generative AI appear to have on intercultural communication through a review of key themes drawn from research, showcasing the proclivity of generative AI to reinforce stereotypes through training data (Dai & Hua, 2025; Jones, 2025), considering its utility for facilitating inclusive education (Passantino, 2024), examining trust and bias in AI-moderated interactions (Jenks, 2025), while urging for collaboration to address ethical considerations. In doing so, we will conclude with a review of two case studies: An investigation into difficulties faced in the detection of edited and AI-generated essays (Wu & Flanagan, 2023) and the role of Gen AI in the ‘cultural transformation’ encountered by female students in an international environment (Song, 2026). The aim of which will be to highlight the tensions between the utility of generative AI programmes versus their potentially inadvertent outcomes, focusing on insights into the way technology can both misinterpret and positively realign intercultural understanding. Overall, the presentation calls attention to ethical integration of AI in intercultural communication research and implementation.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

  1. SDG 4 - Quality Education
    SDG 4 Quality Education

User-Defined Keywords

  • Generative AI
  • intercultural communication
  • algorithmic bias
  • stereotypes
  • ethical considerations
  • cross-cultural dialogue

Fingerprint

Dive into the research topics of 'Generative AI and the Future of Cross-cultural Dialogue?'. Together they form a unique fingerprint.

Cite this