Abstract

This lecture explores the cultural adaptation challenges encountered in the process of human–AI collaborative generation of Tibetan cultural picture books. Focusing on the core tension between “value alignment” (ensuring that AI outputs conform to Tibetan cultural values and aesthetics) and “technical distortion” (algorithm-induced misrepresentation of cultural elements), the study designs and implements a human–AI collaborative picture book generation framework. Through a mixed-methods approach combining eye-tracking experiments and questionnaire surveys, it demonstrates that systematically integrating cultural expertise into the AI generation loop is crucial for mitigating distortion and achieving effective value alignment. The lecture aims to leverage AIGC-related technologies to offer significant theoretical reflections and practical demonstrations for the innovative preservation of intangible cultural heritage.

Speaker Bio

Yuhan Liu, Associate Professor at the School of Design, Huazhong University of Science and Technology, and 2024-2025 visiting scholar at Harvard FAS CAMLab. Her research focuses on digitalization of distinctive ethnic cultures and human-AI interaction.

Event Information

• Free and open to the public
• October 10th (Friday)
• 12:00 – 1:30 PM ET
• This is an in-person event.

Please note

Advanced RSVP required.
The event will be conducted in Mandarin Chinese.