Virginia Tech® home

Large-Scale Human-AI Interactive Curriculum Generation for Online Learning

Researchers will develop a human-artificial intelligence interactive curriculum-generation framework for labor-optimized and trustworthy online learning platforms.

Project funded by: CCI Southwest Virginia Node, in collaboration with Virginia Tech’s Institute for Society, Culture, and Environment and the Tech4Humanity initiative

Project Investigators

Principal Investigator (PI): Dawei Zhou, assistant professor, Virginia Tech Department of Computer Science

Co-PI: Shuo Yao, professor, Radford University School of Communication

Rationale and Background

Online learning platforms provide a fundamental tool for educators to develop, organize, and deliver high-quality courses for students at any place and at any time. 

In the past few years, tremendous efforts have been made to develop a variety of online courses and online learning platforms across a wide range of disciplines. 

While these efforts have been highly successful, questions remain about ways to customize curriculum based on each student’s progress, background, and interest:

  • How can we effectively and efficiently comprehend human interactions (for example, teacher evaluations and student feedback) in online learning platforms to improve the curriculums?
  • How can we secure this digital experience between humans (teachers and students) and online learning platforms to avoid systemic bias and risks? 
  • Given imbalanced student-teacher ratios (thousands of students, one teacher) in online learning platforms, how can we optimize educators’ efforts and time to provide high-quality learning experience to students?

Methodology

To ensure a trustworthy online learning experience, the team will audit and secure the curriculum-generation process by leveraging existing manually developed curriculums and students’ feedback. 

The proposed research framework is composed of three modules, which operate in a mutually beneficial way:

  • Human interaction comprehension: Aims to automatically comprehend human interactions (for example, teacher evaluations and student feedback) to improve online learning platforms.
  • Personalized machine teaching: Generates a personalized curriculum for students by leveraging the existing human-developed curriculum and comprehending the users’ interactions.
  • Curriculum auditing: Allows educators to supervise and audit the curriculum-generation process to ensure a secure online learning experience.

Projected Outcomes

This project will establish new online learning foundations in terms of teaching theories, secure curriculum generation, and human-in-the-loop machine learning. 

The project’s multidisciplinary approach, including deep learning, human-AI intelligence, communication, and teaching theories, will advance individual fields and help accelerate the confluence between them.

Researchers will deliver:

  • An integrated software system.
  • A technical report detailing the novel methods and results.
  • Three paper submissions regarding human interaction comprehension.