Skip to main content Skip to navigation

Generative AI and Its Impact on Personalized Intelligent Tutoring Systems

Project Overview

This document examines the application of generative AI, specifically large language models like GPT-4, within Intelligent Tutoring Systems to personalize and enhance education. The study highlights the use of AI for automated question generation, providing customized feedback, and creating interactive dialogue systems. Key findings reveal potential for improved learning experiences through personalized interaction. However, the report also acknowledges challenges related to pedagogical accuracy, bias in AI outputs, and maintaining learner engagement. Looking ahead, the research explores future directions such as incorporating multimodal AI, developing emotionally intelligent tutoring systems, and addressing ethical implications of AI in education.

Key Applications

Automated Question Generation

Context: Intelligent Tutoring Systems (ITS) adapting to learner's understanding and proficiency level, integrating real-world applications.

Implementation: LLMs like GPT-4 generate questions varying in difficulty, topic specificity, and cognitive demand. These questions can also integrate real-world applications, such as environmental impacts on ecosystems.

Outcomes: Personalized assessment of learner progress, enhanced engagement, deeper cognitive processing, mitigation of question repetition, and making learning more relevant.

Personalized Feedback and Dialogue

Context: Intelligent Tutoring Systems (ITS) and Language learning ITS.

Implementation: LLMs analyze learner input in real-time and generate detailed explanations, hints, or corrective feedback tailored to specific errors or misconceptions. The system engages in increasingly complex conversations as the learner's language skills improve.

Outcomes: Promotes active learning and deeper understanding, enhances engagement and motivation, provides a dynamic and responsive learning environment, and fosters a more personalized and effective learning process.

Implementation Barriers

Pedagogical Accuracy

LLMs may generate inaccurate or misleading educational content due to outdated, incorrect, or biased information in their training data.

Proposed Solutions: Implement robust validation mechanisms, hybrid systems with rule-based guidelines, expert oversight, and continuous monitoring.

Bias and Equity

Generative AI systems inherit biases in their training data, potentially disadvantaging learners from diverse backgrounds.

Proposed Solutions: Use diverse and representative datasets, implement fairness algorithms, conduct regular audits, and involve diverse educators and stakeholders in development and evaluation.

Learner Engagement

Over-reliance on AI-generated content without sufficient human oversight can lead to disengagement.

Proposed Solutions: Incorporate gamification, adaptive learning pathways, varied instructional strategies, and ensure contextually relevant interactions aligned with learner goals.

Project Team

Subhankar Maity

Researcher

Aniket Deroy

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Subhankar Maity, Aniket Deroy

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gemini-2.0-flash-lite