Zhen Xu
Hey there šŸ‘‹
I'm Zhen

I'm a first-year PhD student in Cognitive Science at Teachers College, Columbia University, advised by Prof. Renzhe Yu, where Iā€™m also a member of the AEQUITAS Lab! Before that, I earned my Masterā€™s degrees from both Teachers College (majoring in Learning Analytics) and Beijing Normal University (majoring in STEM Curriculum and Pedagogy), where I was advised by Prof. Ping Zhang.

With a background that connects cognitive science, pedagogy, learning analytics, and AI, Iā€™m interested in understanding how education is being transformed in the age of AI and how we might respond collaboratively.

Recently, I am working on two interconnected directions:

  1. Examining the systemic impact of AI on higher education. I explore how the rise of AI is reshaping educational goals and instructional practices through institutional-level data analysis to capture and understand these shifts at scale.
  2. Evaluating AIED tools from a learning sciences perspective. Iā€™m interested in developing learning science-based evaluation frameworks and approaches for AIED tools, with a focus on their pedagogical capabilities.

Along the way, Iā€™ve also developed a side interest in exploring AI as research tool, exploring how human-AI collaboration can enhance educational research.

šŸ“¢ News

  • 2025.07 ā€“ I will attend AIED, EDM, and L@S in July 2025 in Perlamo, Italy. Happy to connect if you'll be there too!
  • 2025.04 ā€“ I will give an online talk as a Guest Speaker and Panelist at the Educational Technology Sub-Forum of the 16th China Education Symposium at Harvard University.
  • 2025.03 ā€“ I will give an online talk as a Guest Speaker at the AI Application Sub-Forum of Columbia University Teachers College China-US Education Forum.
  • 2025.03 ā€“ I will present a collection of work from our lab at the AI in Education: Research and Practice session, part of the Columbia AI Summit.
  • 2024.11 ā€“ Our paper ā€œIntersectional inequalities in asynchronous academic communicationā€ won Best Student Paper Awards at SCiP 2024šŸ†.
  • 2024.09 ā€“ I will start my PhD in Cognitive Science at Teachers College, Columbia University!

šŸ’” Recent Projects

Pedagogy-Based Evaluation Framework of Virtual Teaching Assistants

Pedagogy-Based Evaluation Framework of Virtual Teaching Assistants
AIED25 Workshop

Siyan Li, Zhen Xu, Vethavikashini Chithrra Raghuram, Xuanming Zhang, Renzhe Yu, Zhou Yu

We developed a pedagogically grounded evaluation framework for Virtual Teaching Assistants (VTAs) in asynchronous forums and identified effective modeling approaches and key generalization challenges.
Evaluating 21st-Century Competencies in Higher-Edu Curricula with LLMs: Benchmarking and Targeted Prompting Strategies

Evaluating 21st-Century Competencies in Higher-Edu Curricula with LLMs: Benchmarking and Targeted Prompting Strategies
Under Review

Zhen Xu, XinGuan, Chenxi Shi, Qinhao Chen, Renzhe Yu

We investigated how LLMs can support 21st-century competency analytics in higher education by benchmarking four models against a manually annotated dataset, and we proposed a prompting strategy (Curricular CoT) that modestly improves accuracy and reduces hallucinations.
From Course to Skill: Evaluating LLM Performance in Curricular Analytics

From Course to Skill: Evaluating LLM Performance in Curricular Analytics
AIED25

Zhen Xu, Xinjin Li, Yingqi Huan, Veronica Minaya, Renzhe Yu

We systematically evaluate LLMsā€™ capabilities on skill extraction task in Curricular Analytics, benchmarking their performance against traditional NLP approaches, and examining how model performance varies across off-the-shelf and task-tailored strategies to offer practical guidance for integrating LLMs into Curricular Analytics workflows.
Last updated: 2025.07 © Zhen Xu