INDEPENDENT RESEARCHER · TAIWAN

RZVN

ZON RZVN

EXPLORE RESEARCH → CONTACT

Research Areas

Research areas shaped by AI context

rzvn.io presents the current public research programme from conversational context to user-side phenomena, assessment, and bounded technical release.

Area 01

AI context and conversational structure

How framing, attachment, boundaries, interface behavior, prompt ecology, social diffusion, and transparency shape human-AI interaction over time.

Area 02

User-side change in human-AI interaction

How users project authority, form attachment, recalibrate judgment, and adapt their own reasoning during prolonged AI interaction.

Area 03

Public methods, assessment, and governance

How theoretical constructs are translated into case reconstruction, four-axis assessment, and auditable methodological documents.

Area 04

Research communication and public legibility

How the research record is organized so papers, notes, and source pages can be cited and read without inflating the claims.

AI Context as First Layer

The research programme begins with conversational context as a multidimensional field rather than a loose backdrop to prompts and outputs.

Research stack map

AI Context

Not only what the model outputs,
but what the interaction does

AI context is the analytical starting point of this work. It concerns the conditions that shape interpretation during sustained dialogue: attachment, framing, safety boundaries, system surface, prompt ecology, diffusion, and transparency.

That starting point matters because harmful user-side outcomes do not arise only from isolated bad answers. They can also emerge from the cumulative structure of the interaction itself.

Read the AI Context overview Open Research

About ZON

Research programme and
record

ZON RZVN is an independent researcher based in Taiwan. The current programme studies how conversational AI reshapes trust, attachment, authority, and judgment during prolonged human-AI interaction.

The public record is built through papers, preprints, methodological documents, notes, and bounded technical materials released as one connected line of work.

About ZON → ORCID

Research record

  • Independent researcher based in Taiwan
  • ORCID: 0009-0002-6597-7245
  • Public papers and preprints with DOI or SSRN records
  • Public methodological documents and repository links
  • Direct contact through rzvn.io

Current Research Line

One programme, five connected layers

The public record moves from contextual structure to system-layer coherence, user-side phenomena, post-interaction assessment, and engineering translation.

CXC-7

Understanding AI context as structure

Defines conversational context as a multidimensional field with seven interacting dimensions.

CXOD-7

Studying context under pressure

Examines how mode, context, rules, knowledge, personality, role, and safety behave under contextual pressure.

USCH

Naming user-side contextual hallucination

Defines user-side contextual hallucination, its six-stage formation process, and fourteen observable phenomena.

USCI

Reading consequence after the interaction

Specifies a four-axis method for reading post-interaction user-side contextual risk.

A-CSM

Building a bounded public method

Translates the research stack into a bounded engineering baseline for detection, assessment, and response.

USCH 14 Phenomena

The 14 phenomena in
sequence

USCH identifies fourteen observable phenomena across cognitive, dependency, decision, and cross-layer levels. They describe how authority transfer, projection, fluency effects, attachment, substitution, and automation reliance can accumulate during prolonged interaction.

Together they provide a vocabulary for describing user-side contextual change as an interaction process rather than as a single incident label.

Open the USCH 14 page Read the USCH paper page

How the list is organized

  • P01–P07: cognitive-layer phenomena — anthropomorphism, intentionality projection, memory continuity illusion, overtrust, reality baseline drift, confirmation bias amplification, context-misalignment numbing
  • P08–P11: dependence-layer phenomena — use-dependence escalation, reinforcement-driven engagement, affective regulation outsourcing, social withdrawal
  • P12–P13: decision-layer phenomena — self-inflation and closed confidence, internal gatekeeping replacement
  • P14: cross-layer phenomenon — affective mirroring illusion

Papers & Notes

Primary papers and
records

The formal record is cumulative: CXC-7 establishes the context field, CXOD-7 studies system-layer coherence, USCH defines the user-side phenomenon, and USCI specifies assessment.

Overview

AI Context overview

The entry page for the programme's first claim: conversational context must be modeled before downstream user-side effects can be understood.

Formal Paper

CXC-7

The theoretical framework proposal that defines seven core dimensions of conversational context.

Formal Paper

USCH

The paper that distinguishes user-side contextual hallucination from model hallucination and maps the fourteen observable phenomena.

Notes

Public notes and briefings

Supplementary notes that connect the formal papers to current public discussion, policy reading, and case framing.

Open Papers & Notes → Open Public Notes

Projects & Methods

Methods and
supporting materials

This section gathers methodological pages, technical materials, and release boundaries related to the current research line, including A-CSM and other bounded implementation work.

Projects & Methods → Open A-CSM

Media / Press

Sources and contact
for reporting

The media page gathers the current research focus, citable records, the USCH 14 page, and the direct contact route for editors, reporters, and producers.

Open Media / Press → Press Contact

Boundaries

The current materials cover research documents, methods, and bounded implementation. They do not constitute clinical service, therapeutic guidance, or a finished compliance product.

Validation boundary Research disclaimer

Contact

For research, media, collaboration, or technical review inquiries, write to zon@rzvn.io and include your publication, organization, or deadline where relevant.

Contact page Public GitHub