1 of 14

Improving College Student’s Engagement and Retention Using Artificial Intelligence

Mohamed Abouzahra

Associate Professor

Analytics and IS

CSUMB

mabouzahra@csumb.edu

G Craig Vachon

Seed Investor, occasional entrepreneur, and student

craig.vachon@gmail.com

2 of 14

Motivation

  • Reduction in the number of students applying to universities

​

  • High rate of dropping off/not finishing in five years

​

  • Lack of classroom engagement

​

3 of 14

Extra motivation

  • Students do not graduate with enough job market skills

​

  • Continuous changes in job market skills

​

  • Self-training in organizations is not engaging

​

4 of 14

Hypothesized Solution

  • Create a digital twin for each student:
    • and tailor content to fit the student’s needs;
    • and provide a teaching methodology/delivery mechanism that best meets the student’s needs (most effective methodology)
  • Creating a student-centered learning environment (while providing a simple interface for the educator)

​

5 of 14

Can AI help to implement this solution?

  • GenAI models allow the creation of new content based on previous knowledge
  • GPT 3 and GPT 4 base models have been used to create text content
  • GPT 4o can create different types of content including images/video

​

6 of 14

System Components

  • Student digital twin
    • Database of student information related to their learning styles and objectives
  • Content Creator
    • AI system that is trained/fine-tuned on content and will generate dynamic training modules
  • Evaluator
    • Assesses and evaluates student performance
    • Integrates with the content creator to generate new training
    • Updates Student digital twin

7 of 14

How the system works

8 of 14

Benefits

  • Better Student Engagement

​

  • Can be applied in hybrid/fully online settings

​

  • Can be updated with new job-market skills

​

  • Can be extended to alumni

9 of 14

Scenario

  • Teaching an AI course

10 of 14

Functions as a comic

11 of 14

Functions as an interactive activity

12 of 14

Challenges

  • Getting the data

​

​

  • Training the model

​

​

  • Privacy

13 of 14

Using Web 3.0 technology

14 of 14

Questions?