Advances in generative AI, speech synthesis, and embodied avatars enable systems that not only assist communication, but can act as proxies on users' behalf. Prior work in HCI has largely focused on systems as external tools, with less attention paid to the experiential consequences of users' speech and actions becoming assimilated with AI-generated output. We introduce the design and implementation of ProxyMe, a work-in-progress VR prototype that allows users to embody an avatar whose voice and spoken content are modified by an AI system. By combining avatar-based embodiment, voice cloning, and AI-mediated speech augmentation, ProxyMe invites the exploration of avatar self-extension: situations in which AI-modified communication is experienced as part of one's own expressive behavior. We chart out research challenges and envisioned scenarios, with a focus on how varying degrees of delegation and steerability can influence perceived agency, authorship, and self-identification.

翻译：生成式人工智能、语音合成与具身化化身技术的进步，使得系统不仅能够辅助沟通，更能作为用户在虚拟空间中的代理。以往人机交互研究主要将系统视为外部工具，较少关注用户言行与AI生成内容相融合所带来的体验性后果。本文介绍了进行中的VR原型系统ProxyMe的设计与实现，该系统允许用户操控一个其语音及说话内容经AI系统修改的化身。通过整合基于化身的具身体验、语音克隆与AI介导的语音增强技术，ProxyMe旨在探索化身自我延伸现象：即当经AI修改的交流行为被体验为个体自身表达行为组成部分的情境。我们系统梳理了研究挑战与前瞻场景，重点关注不同程度的代理权限与可操控性如何影响用户对能动性、创作权及自我认同的感知。