Skip to content
View fengchiwei's full-sized avatar

Block or report fengchiwei

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
fengchiwei/README.md

Chi-Wei Feng (馮奇偉)

Independent Researcher in Educational Technology & Human-AI Co-Construction
Taipei, Taiwan | Contact: fengchiwei@gmail.com


🔬 About & Research Interests

I am an educator and independent researcher based in Taiwan, focusing on the intersection of Inquiry-Based Science Education, Epistemic Agency, and Human-AI Interaction.

My core research interests include:

  • Scientific Epistemological Views (科學認識觀): How learners evaluate the source, certainty, development, and justification of knowledge.
  • Learners' Epistemic Justification in AI Interaction (人機互動中知識求證機制): How students critically evaluate and verify responses generated by Large Language Models (LLMs) rather than treating AI as an unquestioned answer dispenser.
  • Anti-Sycophancy & Pedagogical Scaffolding in LLMs (大語言模型防迎合與教育鷹架): Mitigating LLM sycophancy (the tendency of models to validate incorrect student statements or conform to misconceptions) and transforming conversational AI from direct answer-providers into inquiry-guiding coaches.
  • Misconception Diagnosis (迷思概念診斷): Designing structured dialogue hints that guide learners to identify contradictions in their own reasoning.

🚀 Current Research (In Preparation)

Investigating Sycophancy and Epistemic Agency in Conversational Agents

  • Focus: Controlled benchmark testing of language models using classic science misconception problem sets.
  • Goal: Evaluating whether models hold, abandon, or appropriately revise a correct scientific position when a simulated learner asserts a misconception, and whether hint-based prompting helps learners reason instead of handing them the answer, in Traditional Chinese and English.
  • Methodology: Controlled simulations with standardized misconception probes and a coding rubric. No human subjects are involved in Phase 1.
  • Outputs: The probe set and coding rubric will be released openly.

🎓 Academic Background & Past Research

Master of Education (M.Ed.) Thesis (2010)

  • Institution: Department of Natural Sciences, Taipei Municipal University of Education (臺北市立教育大學自然科學系教學碩士班)
  • Advisor: Dr. Mei-Fen Wang (王美芬 博士)
  • Degree Conferred: July 2010
  • Thesis Title:
    The Effects of Reading Science News Texts on Sixth Graders' Scientific Epistemological Views and Attitudes toward Science
    (《閱讀科學新聞文本對國小六年級學童科學認識觀與對科學態度的影響》)

Abstract & Key Findings

This study investigated the effects of a two-month science news reading activity on sixth graders' (N = 30) scientific epistemological views and attitudes toward science. It was a quantitative study combined with qualitative data.

  • Scientific epistemological views: Posttest scores were significantly higher than pretest, with significant differences in the dimensions of source (來源) and justification (求證), but not certainty and development.
  • Attitudes toward science: Posttest scores were significantly higher than pretest in the dimension of attitude toward science-self, but not in attitude toward attending scientific investigation activities or the viewpoint of scientists and scientific careers. There was no significant correlation between epistemological views and attitudes toward science.
  • Students' views of the activity: The qualitative study showed most of the students have positive attitude about reading scientific news. Through the activity, more knowledge can be learned, possible applications in the daily life, and the scientific news are interesting were the main reasons.
  • Significance for Current Work: Provides the empirical and methodological foundation for extending epistemological inquiry from science news texts (2010) to AI-generated texts and conversational agents (2026).

📬 Contact & Collaboration

I welcome academic discussions, potential research collaborations, and scholarly feedback on educational AI ethics, prompt scaffolding, and epistemic evaluation.

  • Email: fengchiwei@gmail.com
  • Location: Taipei, Taiwan

Pinned Loading

  1. build-livecd build-livecd Public

    Automatically exported from code.google.com/p/build-livecd

    JavaScript

  2. codextest codextest Public

  3. git-credential-manager git-credential-manager Public

    Forked from git-ecosystem/git-credential-manager

    Secure, cross-platform Git credential storage with authentication to GitHub, Azure Repos, and other popular Git hosting services.

    C#

  4. haxm haxm Public

    Forked from intel/haxm

    Intel® Hardware Accelerated Execution Manager (Intel® HAXM)

    C

  5. https-github.com-mirage335-IQEmu https-github.com-mirage335-IQEmu Public