Researchers

TAKECHI Kaoruko

TAKECHI Kaoruko
Associate Professor
Faculty Department of Business and Economics
Researchmap https://researchmap.jp/0202

Research Activities

Research Areas

  • Informatics , Human interfaces and interactions
  • Humanities & social sciences , Linguistics
  • Humanities & social sciences , Educational technology
  • Humanities & social sciences , Foreign language education
  • Humanities & social sciences , Educational psychology

Research Interests

Generative AI in Education, Muti-agent deliberation / orchestration, Outlier quarantee / dissent, Evidence-linked audit logs / auditability, Vocabulary acquisition ＆assessment, English for Specific Purposes, Metacognition / SRL

Published Papers

  1. Who Is Deciding Here? A Five-Layer Model of Delegation-Boundary Crossing in a Single-LLM Multi-Persona Configuration
    Kaoruko Takechi
    Proceedings of the 14th International Conference on Human-Agent Interaction (HAI '26)  Nov. 2026  , Refereed
  2. Proposal of the Reflection-Completion Model (RC-Model): A Metacognition-Inducing Discussion Support System Using Generative AI
    Kaoruko Takechi
    The Bulletin of Kindai University Junior College Division  58  (2)  , 45-67, 31, Jan. 2026 
  3. Combined Use of Zoom and Google Form in English Lessons
    Kaoruko Takechi
    近畿大学医学部 教育センター紀要  1  (1)  , 21-24, Jan. 2022 

MORE

Conference Activities & Talks

  1. Who Is Deciding Here? A Five-Layer Model of Delegation-Boundary Crossing in a Single-LLM Multi-Persona Configuration , Kaoruko Takechi , 14th International Conference on Human-Agent Interaction (HAI 2026) [Poster accepted; presentation forthcoming] , Nov. 2026
  2. 議論支援アプリ「思考アスレチック」の設計とHAI展開への道 ― 非線形分岐型思考支援の考察 , 武知 薫子; 山根 愼之輔; 植田 素直; 大谷 慈輝; 福田 尚己; 立石 楓 , HAIシンポジウム2026 , 28, Feb. 2026
  3. Illuminating Blind Spots: A Metacognitive Multi―Agent Framework for Robust Argumentation , Kaoruko Takechi , The Japan Association for College of Technology , 5, Sep. 2025

MORE

MISC

  1. Who Is Deciding Here? Delegation-Boundary Drift in a Single-LLM Multi-Persona Configuration , Kaoruko Takechi , Zenodo , 19, Apr. 2026
    Summary:Preprint, version 1. An exploratory longitudinal case study of delegation-boundary drift in a single-LLM multi-persona configuration, based on 16 months of autobiographical use. Examines opacity of speaker attribution, short-circuiting of the chain of command, and substitution of judgment, and discusses the design implications of defining the coordinator relationally.
  2. Scope Lock: When the LLM Restrains Itself , Kaoruko Takechi , Zenodo , 8, Mar. 2026
    Summary:Preprint, version 3. Reports Scope Lock, a conversation protocol combining role fixation, purpose declaration, and prohibition clauses to suppress interference patterns in LLM dialogue. Examines partial suppression and stability bounded by persona cardinality.
  3. Cross-Lingual Blind Spot: Why Japanese Users See "Victimized AI" When They Say "Atsukau" , Kaoruko Takechi , Zenodo , 29, Jan. 2026
    Summary:This short paper reports an observation on meaning shifts that occur in text-to-image AI systems accepting Japanese input. A prompt that gained popularity among Japanese users in early 2026, 「これまで私があなたをどう扱ってきたのか画像にしてください」("Please create an image showing how I have treated you until now"), produced markedly different outputs depending on how the Japanese verb "atsukau" ("to handle/deal with") was interpreted in English. When the system simplistically translated "atsukau" as "treat", the English term's nuances of manner and evaluation activated patterns depicting AI as a sentient victim. This paper documents the phenomenon across two models (GPT and Gemini), discusses the linguistic mechanism, and proposes the concept of "cross-lingual blind spot": the likelihood that users are unaware their native-language prompts undergo semantic transformation through English-language internal processing.

MORE

Research Grants & Projects

  1. Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research, Analysis of the Anatomy of Fear of Failure in Studying Overseas: Development of A Cross-Cultural Game to Control the Fear , Kindai University
  2. 独立行政法人日本学術振興会, 科学研究費助成事業, グローバル人材育成を目指した大学留学プログラム構築：韓国長期留学の質的観察から
  3. 近畿大学, 学内研究助成金（21世紀教育開発奨励金）, グローバル人材育成を目指す留学プログラムの構築ー近畿大学生は留学を通して何をどのように学んだのか

MORE