Avatar beyond replication: digital human thought twins as co-creators through AI mind design in Eastern and Western thought
摘要
As artificial intelligence increasingly mediates artistic expression, this paper investigates whether AI-based digital human twins can function as co-creative partners beyond replication-oriented models. It traces the shift from digital avatars to Digital Human Thought Twins in artistic contexts, drawing on Western philosophical discussions of agency (Kant, Sartre, Heidegger) and Eastern concepts of relational emergence (Buddhist śūnyatā and Daoist yin–yang). Through analysis of cross-cultural case studies—including An Evening with Whitney, Holly + , PLAVE, VNCCII, and Me and My Digital Twin—this study engages phenomenology and cognitive science to reconsider established assumptions about authorship and creativity. Project SYN-METHEUS serves as an illustrative framework for examining human–AI co-creation. It explores three speculative scenarios: Creative Extension through Automation, Self-Extension via Retrospective Digital Twins, and Creative Domain Expansion through Collaborative Creation. At the center of this project is Higgs, an AI-powered Digital Human Thought Twin developed with GPT-4o. Higgs implements the A.R.T. Process (Analyzing—Reflecting—Turn-taking), an AI Mind Design framework that supports real-time co-creative interaction with human creators and audiences. Future development may incorporate Retrieval-Augmented Generation (RAG) to support greater contextual grounding and continuity. An exploratory assessment using a modified User Experience Questionnaire (UEQ) indicated strengths in Autonomous Learning & Evolution (2.67/3.0) and Interactivity & Responsiveness (2.50/3.0), alongside limitations in Environmental Adaptability in multi-user contexts. These findings suggest that, within this experimental setting, Digital Human Thought Twins can be understood as reflective co-creative partners in artistic practice. More broadly, the study contributes to ongoing discussions of artistic identity, authorship, and creativity in AI-mediated environments.