NSF ER2: Responsible AI for Data-Intensive Educational Research

Developing a practical framework for responsible use of artificial intelligence in data-intensive educational research.

Overview

Data-intensive education research uses information about student learning, including demographic information, academic performance, behavior, attendance, survey responses, and technology use. Artificial intelligence can help researchers identify patterns, predict performance, identify students who may need support, and personalize learning. It can also raise ethical concerns, expose sensitive information, and contribute to harmful decisions.

This project develops a practical framework for responsible AI use in education research, with attention to fairness, privacy, transparency, and stakeholder needs. The work connects ethical principles with tools, communication resources, and educational materials that support informed decisions about education data.

Research approach

The project brings together three stakeholder perspectives:

  • Data administrators, who manage access to education data.
  • Researchers, who use education data for analysis.
  • Participants, including parents and older students who decide whether to take part in research studies.

The research program combines stakeholder studies, responsible AI methods, communication design, and educational materials.

1. Understand stakeholder needs

Surveys, interviews, and focus groups will examine stakeholder practices, concerns, requirements, and decisions related to the use and sharing of education data.

2. Assess fairness and privacy risks

The project will develop a user-oriented platform to help identify potential bias and privacy risks in datasets and models.

3. Explain AI risks and benefits

A toolkit will help education researchers communicate with data administrators and participants. It will translate technical concepts, model limitations, data requirements, fairness concerns, and privacy risks into accessible explanations and resources.

4. Support learning and practice

Educational materials will be tailored to stakeholder needs.

Project team

  • Yongkai Wu, Clemson University - PI
  • Chiu C. Tan, Temple University - Site-PI
  • Ting Sun, University of Utah - Site-PI
  • Yucong Dai, Clemson University - Graduate Research Assistant

Funding

This project is supported by the U.S. National Science Foundation under the following awards: 2520496, 2520497, and 2520498.