Skip to main content Skip to navigation

Exploring Student Behaviors and Motivations using AI TAs with Optional Guardrails

Project Overview

The document examines the integration of AI-powered teaching assistants (TAs) in programming education, emphasizing a study that investigates student behaviors and motivations in relation to an AI TA equipped with optional guardrails. It highlights the advantages of AI TAs, such as delivering timely and personalized feedback, which can enhance the learning experience. However, it also raises concerns about potential over-reliance on these tools and the risk of academic misconduct. The study reveals that a significant number of students frequently use a 'See Solution' feature, especially under time constraints, which is linked to decreased performance. These findings underscore the importance of aligning pedagogical objectives with student needs when developing AI TAs to foster effective educational outcomes.

Key Applications

Edugator Tool with AI TA

Context: Large introductory programming course for university students enrolled in programming classes.

Implementation: The AI TA offers optional guardrails and a 'See Solution' feature during coding tasks.

Outcomes: Students receive immediate feedback, with varied usage of the 'See Solution' feature based on performance level.

Challenges: Students may over-rely on the tool, which can hinder their learning and problem-solving skills.

Implementation Barriers

Educational

Concerns about student over-reliance on AI tools, limiting development of independent problem-solving skills.

Proposed Solutions: Incorporating guardrails in AI TAs to encourage self-directed learning and problem-solving.

Technical

Inaccuracies in AI-generated content which may lead to misconceptions.

Proposed Solutions: Providing instructor oversight on AI outputs to ensure pedagogical soundness.

Behavioral

Students procrastinating and relying on AI solutions close to deadlines.

Proposed Solutions: Encouraging timely engagement with course materials and promoting self-regulation skills.

Project Team

Amanpreet Kapoor

Researcher

Marc Diaz

Researcher

Stephen MacNeil

Researcher

Leo Porter

Researcher

Paul Denny

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Amanpreet Kapoor, Marc Diaz, Stephen MacNeil, Leo Porter, Paul Denny

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