Do Students Using Cursor Achieve Higher Learning Gains in Algorithm Design Compared to Those Using Traditional IDEs? A Randomized Controlled Trial
Year: 2025
Type: article
Study designs: Randomized Controlled Trial, Clinical Trial
Abstract: Background: AI-powered integrated development environments (IDEs) are rapidly transforming programming education. While passive AI tools like GitHub Copilot have been studied, the educational impact of agentic AI IDEs like Cursor remains unexplored. Objective: This study investigates whether undergraduate computer science students using Cursor (an agentic AI IDE) demonstrate significantly higher learning gains in algorithm design compared to students using traditional IDEs without AI assistance. Methods: We conducted a randomized controlled trial with 48 Gen Z undergraduate CS students over 4 weeks. Participants were randomly assigned to use either Cursor (experimental group, n=24) or VS Code without AI extensions (control group, n=24). Learning gains were measured through pre/post knowledge tests, assignment performance, debugging efficiency, and self-efficacy changes. Data were analyzed using independent samples t-tests and correlation analyses. Results: The Cursor group demonstrated significantly higher learning gains (M=19.7, SD=8.8) compared to the VS Code group (M=12.8, SD=9.2), t(46)=2.71, p=.009, Cohen’s d=0.78. The Cursor group also showed superior performance in assignment scores (M=66.9 vs M=64.8, p=.042), debugging efficiency (p<.001), and selfefficacy changes (M=14.1 vs M=8.3, p<.001). Qualitative analysis revealed that while Cursor users appreciated the learning support, some expressed concerns about AI dependency. Conclusions: Agentic AI IDEs like Cursor can significantly enhance learning gains in algorithm design education. The mediumto-large effect size suggests that AI-powered tools provide effective scaffolding that reduces cognitive load on syntax and debugging, allowing students to focus on algorithmic thinking. However, educators should implement strategies to prevent over-reliance and ensure students develop fundamental programming skills alongside AI proficiency. Keywords: Agentic AI, Programming Education, Algorithm Design, Cursor IDE, Learning Gains, Gen Z Students, Computer Science Education, AI-Assisted Learning
Author: Yaeesh, Baker
Institution: An-Najah National University
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7106574353
FWCI:
Cites:
Cited by:
Related to: