Skip to main content Skip to navigation

Improving Student Learning with Hybrid Human-AI Tutoring: A Three-Study Quasi-Experimental Investigation

Project Overview

The document presents a quasi-experimental study investigating the efficacy of a hybrid human-AI tutoring model designed to enhance math learning outcomes for middle school students, particularly those from low-income backgrounds. It emphasizes the transformative potential of AI-assisted tutoring in fostering engagement and improving learning experiences, especially for students who struggle academically. The study reveals significant positive effects on both learning processes and overall student achievement, underscoring the role of generative AI in personalized education. Additionally, it addresses critical challenges related to equitable access to quality educational resources, advocating for further research and enhancements in implementation strategies to maximize the benefits of AI in educational settings. Overall, the findings suggest that integrating generative AI in tutoring can be a powerful tool for improving educational outcomes, particularly for disadvantaged students, while also highlighting the need for ongoing evaluation and adaptation in the deployment of such technologies.

Key Applications

Hybrid Human-AI Tutoring

Context: Middle school students from low-income backgrounds in urban schools in the U.S.

Implementation: A three-study quasi-experimental design involving human-AI tutoring across three different sites, utilizing various math software and tutor-student interactions.

Outcomes: Positive effects on student proficiency and engagement, particularly benefiting lower-achieving students compared to their higher-achieving peers.

Challenges: Access to technology and digital devices, variability in implementation across different school sites, and the potential for selection bias in study design.

Implementation Barriers

Access Barrier

Limited access to technology and digital devices among students from low-income backgrounds.

Proposed Solutions: Collaborative efforts to ensure equitable access to necessary technological resources.

Implementation Barrier

Variability in implementation fidelity across different school sites affecting overall intervention effectiveness.

Proposed Solutions: Standardizing implementation protocols and monitoring fidelity to ensure consistent application of the tutoring model.

Research Barrier

Challenges in measuring long-term effectiveness and outcomes due to the need for extensive data collection and analysis.

Proposed Solutions: Utilizing rapid-cycle quasi-experimental designs to gather preliminary data while planning for more rigorous evaluations.

Project Team

Danielle R. Thomas

Researcher

Jionghao Lin

Researcher

Erin Gatz

Researcher

Ashish Gurung

Researcher

Shivang Gupta

Researcher

Kole Norberg

Researcher

Stephen E. Fancsali

Researcher

Vincent Aleven

Researcher

Lee Branstetter

Researcher

Emma Brunskill

Researcher

Kenneth R. Koedinger

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Danielle R. Thomas, Jionghao Lin, Erin Gatz, Ashish Gurung, Shivang Gupta, Kole Norberg, Stephen E. Fancsali, Vincent Aleven, Lee Branstetter, Emma Brunskill, Kenneth R. Koedinger

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies