Work

Hi, past-human: A multimodal AI-driven dialogue system for triggering more-than-human thinking

Year: 2025
Type: dissertation
Abstract: Despite living in a complex and high-tech era, human-centered thinking continues to dominate how we see and change the world. In recent years, the emerging more-than-human theory has introduced new possibilities for imagining the future. It emphasizes the interdependent relationships among humans, non-human lifeforms, technology, and the environment to foster coexistence, diversity, and alternative worldviews. This theory has already supported various disciplines in exploring more-than-human perspectives and envisioning a future in which humans and others coexist. However, in the field of generative AI, most remain designed based on human-centered purposes, with only limited exploration into whether AI can help users step outside human-centered views and explore more-than-human perceptions. In specific, the potential for generative AI to trigger decentering and cognitive shifting—psychological processes that help people move beyond existing fixed perceptions and explore unfamiliar ways of thinking—remains underexplored. This research responds to that gap by presenting \textit{HI, PAST-HUMAN}, a multimodal AI-driven dialogue system grounded in more-than-human theory. The system uses AI to simulate speculative more-than-human entities from a future world and uses text, sound, image, and video generation to create unpredictable, role-diverse interactions. A user study was conducted with six participants experienced in creative practices, using a combination of think-aloud protocols and semi-structured interviews to examine cognitive shifting, decentering, and emotional responses during the interaction. Through thematic analysis, the study analyzed key moments when participants experienced decentering and cognitive shifts. The research findings show that the system, through its equal dialogue approach and unique, unexpected character design and multi-modal expressions, can effectively encourage users to reflect on the relationships between humans and non-human life forms, and to imagine a more-than-human future. This provides preliminary evidence for the potential of generative AI to support more-than-human design thinking. At the same time, the study highlights current limitations in areas such as character responsiveness and co-creative dynamics, pointing toward opportunities for future interdisciplinary development.
Author: Ge, Xuemei
Institution: Aalto University
Language: English
Open Access status: green
OpenAlex ID: W7078796415
Cites: 49
Cited by: 1
Related to: 10