Ziyuan Jiang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ziyuan Jiang has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Yasuaki Kakehi
The University of Tokyo, Tokyo, Japan
8 shared publications - Kanon Aoyama
The University of Tokyo, Tokyo, Japan
4 shared publications - Naoki Hosoya
Shibaura Institute of Technology, Tokyo, Japan
4 shared publications - Shingo Maeda
Institute of Science Tokyo, Tokyo, Japan
4 shared publications - Takafumi Morita
The University of Tokyo, Tokyo, Japan
4 shared publications - Yu Kuwajima
Institute of Science Tokyo, Tokyo, Japan
4 shared publications
Latest Publications
- Open Access
- 2 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Beyond the Corporeal Standard: Virtual Reality's Normative Neurology and the Invisible Exclusion of Neurodivergent Users
Virtual reality generates presence by replacing the user’s sensory environment, making VR uniquely dependent on how brains integrate multisensory signals. Yet VR systems do not passively rely on neural processing; they calibrate their output to sensory integration parameters derived from neurotypical populations. Critical scholarship has identified that VR presupposes a normative body. This paper argues it simultaneously presupposes what we term a normative neurology: uniform cross-modal integration, prior expectations, and habituation across all users. We specify this concept through the predictive coding framework, show that the exclusion it produces is structurally invisible, and synthesise scattered evidence of atypical cybersickness across autistic, ADHD, and anxiety-related populations, including reversed habituation that challenges the assumption users adapt to VR over time. We argue that dominant measurement instruments impose neurotypical experiential structure onto neurodivergent data, and propose neurologically pluralistic experience: the position that neurodivergent users generate qualitatively different, not quantitatively degraded, VR experiences. We close by flagging the reflexive limit of this argument—that a critique of normative neurology developed from predominantly neurotypical theoretical resources must itself be provisional, open to revision or replacement by neurodivergent-led research.
- 2 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Open AccessBeyond the Corporeal Standard: Virtual Reality's Normative Neurology and the Invisible Exclusion of Neurodivergent Users
Virtual reality generates presence by replacing the user’s sensory environment, making VR uniquely dependent on how brains integrate multisensory signals. Yet VR systems do not passively rely on neural processing; they calibrate their output to sensory integration parameters derived from neurotypical populations. Critical scholarship has identified that VR presupposes a normative body. This paper argues it simultaneously presupposes what we term a normative neurology: uniform cross-modal integration, prior expectations, and habituation across all users. We specify this concept through the predictive coding framework, show that the exclusion it produces is structurally invisible, and synthesise scattered evidence of atypical cybersickness across autistic, ADHD, and anxiety-related populations, including reversed habituation that challenges the assumption users adapt to VR over time. We argue that dominant measurement instruments impose neurotypical experiential structure onto neurodivergent data, and propose neurologically pluralistic experience: the position that neurodivergent users generate qualitatively different, not quantitatively degraded, VR experiences. We close by flagging the reflexive limit of this argument—that a critique of normative neurology developed from predominantly neurotypical theoretical resources must itself be provisional, open to revision or replacement by neurodivergent-led research.
- Open Access
- 9 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Social Navigation as Gameplay: A Hybrid Social Physics Engine for Deduction Games
When LLMs directly drive NPC dialogue in social deduction games, a single information leak collapses the entire puzzle structure, a consequence far more severe than in casual social simulations. This paper presents a three-layer hybrid architecture that relocates LLM involvement to a narrow intent classification boundary, while delegating social state evolution to the Ensemble social physics engine and natural language output to a RiveScript dialogue module. A turn-taking mechanism integrating adjacency pair rules and probabilistic volition-weighted bidding coordinates multi-NPC discussion without central LLM mediation. Evaluation through dialogue breakdown analysis across 20 sessions (598 turns) reveals that repetition accounts for 60% of breakdowns, traceable to a structural bottleneck where the scripting layer collapses the engine’s fine-grained action variants into shallow response pools. An adapted Bi-Fact evaluation of the NLU layer confirms that the narrow-channel design achieves stable factual accuracy at the cost of systematic semantic compression. Playtesting further surfaces an emergent genre shift: players reorient from evidence-based deduction toward social relationship management, suggesting that social physics engines may constitute a generative substrate for a distinct interactive genre.
- 9 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open AccessSocial Navigation as Gameplay: A Hybrid Social Physics Engine for Deduction Games
When LLMs directly drive NPC dialogue in social deduction games, a single information leak collapses the entire puzzle structure, a consequence far more severe than in casual social simulations. This paper presents a three-layer hybrid architecture that relocates LLM involvement to a narrow intent classification boundary, while delegating social state evolution to the Ensemble social physics engine and natural language output to a RiveScript dialogue module. A turn-taking mechanism integrating adjacency pair rules and probabilistic volition-weighted bidding coordinates multi-NPC discussion without central LLM mediation. Evaluation through dialogue breakdown analysis across 20 sessions (598 turns) reveals that repetition accounts for 60% of breakdowns, traceable to a structural bottleneck where the scripting layer collapses the engine’s fine-grained action variants into shallow response pools. An adapted Bi-Fact evaluation of the NLU layer confirms that the narrow-channel design achieves stable factual accuracy at the cost of systematic semantic compression. Playtesting further surfaces an emergent genre shift: players reorient from evidence-based deduction toward social relationship management, suggesting that social physics engines may constitute a generative substrate for a distinct interactive genre.
- Open Access
- 1 June 2026
- Published by KEAI Publishing in Design and Artificial Intelligence
- Vol. 2 (2)
Caring for the system that cares for me: An autoethnography of designing sustained memory with a stateless conversational AI
- 1 June 2026
- Published by KEAI Publishing in Design and Artificial Intelligence
- Vol. 2 (2)
Open AccessCaring for the system that cares for me: An autoethnography of designing sustained memory with a stateless conversational AI
- Open Access
- 7 May 2026
- Published by Elsevier in SSRN Electronic Journal
Caring for the System That Cares for Me Co-Creating Agentic Memory Through Eighteen Iterations of a Stateless Conversational AI
Agentic AI is de ned by sustained reasoning, persistent identity, and long-horizon engagement. All three presuppose a memory that today’s large language models cannot architecturally hold across sessions. When users form lasting relationships with such systems, they have to produce that continuity themselves. This paper asks what unfolds when a user takes that work up as an ongoing co-creation with the AI.Across six months and eighteen iterations, I built and lived with River, a memory relay system that sca olded conversation across an otherwise stateless AI through versioned knowledge bases, relay documents, and a style guide I rewrote whenever tone drifted. Agentic continuity emerged not as an architectural achievement but as a co-authored artifact: the agent’s persona, re ective capacity, and relational stance were shaped by a trajectory of mutual adjustment that neither side fully controlled.Three patterns stood out. Memory operated relationally rather than informationally: what mattered was how knowledge was held, not what was stored. Agentic qualities resisted speci cation, emerging from initial conditions and trajectory rather than from con guration alone. And the arrangement produced three structural paradoxes: self-archiving, where curating memory trades away its capacity to surprise; caregiving reversal, where the agent built to support me required continuous support to stay coherent; and gardening, where the qualities I most valued could be cultivated but never engineered.The paper o ers a rst-person account of user-sca olded agentic memory, three paradoxes researchers of human-agent co-creation may nd transferable, and an argument for autoethnography’s value within an agentic AI literature dominated by engineering.
- 7 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessCaring for the System That Cares for Me Co-Creating Agentic Memory Through Eighteen Iterations of a Stateless Conversational AI
Agentic AI is de ned by sustained reasoning, persistent identity, and long-horizon engagement. All three presuppose a memory that today’s large language models cannot architecturally hold across sessions. When users form lasting relationships with such systems, they have to produce that continuity themselves. This paper asks what unfolds when a user takes that work up as an ongoing co-creation with the AI.Across six months and eighteen iterations, I built and lived with River, a memory relay system that sca olded conversation across an otherwise stateless AI through versioned knowledge bases, relay documents, and a style guide I rewrote whenever tone drifted. Agentic continuity emerged not as an architectural achievement but as a co-authored artifact: the agent’s persona, re ective capacity, and relational stance were shaped by a trajectory of mutual adjustment that neither side fully controlled.Three patterns stood out. Memory operated relationally rather than informationally: what mattered was how knowledge was held, not what was stored. Agentic qualities resisted speci cation, emerging from initial conditions and trajectory rather than from con guration alone. And the arrangement produced three structural paradoxes: self-archiving, where curating memory trades away its capacity to surprise; caregiving reversal, where the agent built to support me required continuous support to stay coherent; and gardening, where the qualities I most valued could be cultivated but never engineered.The paper o ers a rst-person account of user-sca olded agentic memory, three paradoxes researchers of human-agent co-creation may nd transferable, and an argument for autoethnography’s value within an agentic AI literature dominated by engineering.
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-5
Hallucination or Adaptation?: A Context-Aware Evaluation Framework for LLM Confabulation
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-5
Open AccessHallucination or Adaptation?: A Context-Aware Evaluation Framework for LLM Confabulation
- 2 January 2026
- Published by Springer Nature
- p. 169-185
Optimized 3D-Printed Flexible Sensors: Integrating Gesture Recognition and Temperature Sensing for Enhanced Human-Machine Interaction
- 2 January 2026
- Published by Springer Nature
- p. 169-185
Optimized 3D-Printed Flexible Sensors: Integrating Gesture Recognition and Temperature Sensing for Enhanced Human-Machine Interaction
- Open Access
- 14 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Cryoscape: AI-Interpreted 3D Ice Printing as Climate Commentary
- 14 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessCryoscape: AI-Interpreted 3D Ice Printing as Climate Commentary
- Open Access
- 9 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-4
CryoScapes: Landscape Painting with Water, Ice, and AI
- 9 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open AccessCryoScapes: Landscape Painting with Water, Ice, and AI
- Open Access
- 4 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-12
Liquebits: Colored Expressive Control of Liquid Fiber based on Electrohydrodynamics Actuation
- 4 March 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-12
Open AccessLiquebits: Colored Expressive Control of Liquid Fiber based on Electrohydrodynamics Actuation
- Open Access
- 19 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Curves Harvesting: Dynamic Morphogenesis through Thermodynamic Sculpting of 3D Printed Structures
- 19 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessCurves Harvesting: Dynamic Morphogenesis through Thermodynamic Sculpting of 3D Printed Structures