Analisis Efektivitas Penggunaan AI Generatif (ChatGPT) Dalam Pengembangan Perangkat Lunak: Sistem Komparasi Produktivitas Developer Mahasiswa Manajemen Informatika

Abstract

The rapid adoption of Generative AI, particularly ChatGPT, has transformed software development practices globally. While prior studies have examined ChatGPT’s technical capabilities in code generation and debugging, there remains limited empirical understanding of how different user groups—novice developers and management informatics students—actually leverage this tool in real-world software development tasks. This study aims to analyze the effectiveness of ChatGPT usage in software development processes by comparing the productivity of novice developers and management informatics students. A quantitative experimental approach with a comparative design was employed, involving 40 participants divided into two groups: 20 novice developers and 20 management informatics students. Each participant completed identical software development tasks (debugging, code generation, and code explanation) under two conditions: with ChatGPT assistance and without ChatGPT assistance. Data were collected through task completion time, accuracy scores, and a Likert-scale questionnaire. The results show that ChatGPT usage significantly improves productivity for both groups, with an average time reduction of 42% for novice developers and 38% for management informatics students. However, a quality paradox was identified: while speed increased, participants required substantial verification efforts to ensure code correctness. The study concludes that ChatGPT is an effective productivity tool for software development, but its integration requires careful consideration of verification practices and critical thinking skills.