AI-generated future selves have been explored for self-reflection and counseling, yet most remain confined to chat, voice, or diary formats. We introduce RoF (Reminiscences of Futures), a system that enables users to encounter and converse with embodied future selves. It synthesizes voice and appearance from personal data and generates three projections of the user ten years ahead, situated within a tunnel-like installation that evokes a time-slip experience. We report the design process, stepwise judgements, and a formative study with six parent–child pairs that informed design strategies. We also present a week-long user study with fifteen participants. Participants perceived RoF as an emotionally rich experience, reporting shifts in their attitudes toward the future and deeper reflection on their present identity and life. We discuss the implications of interacting with future selves for fostering reflection and preparing positively for the future, as well as ethical issues, such as self-replication using AI.
ACM CHI Conference on Human Factors in Computing Systems