Automating software testing using artificial intelligence and machine learning
TL;DR
Evaluating the effectiveness of AI- and ML-assisted software test automation versus a conventional approach in Software Technology Engineering students at the Universidad Autónoma de Nuevo León, Mexico found the experimental group reduced the mean resolution time and detected a greater number of defects.
Abstract
Introduction: Software test automation is an essential strategy to reduce times, improve coverage and detect defects during development. The incorporation of artificial intelligence (AI) and machine learning (ML) opens up new possibilities for generating, prioritizing, and optimizing test cases. Objective: To evaluate the effectiveness of AI- and ML-assisted software test automation versus a conventional approach in Software Technology Engineering students at the Universidad Autónoma de Nuevo León, Mexico. Methodology: A quantitative, applied, experimental, prospective and comparative study was carried out with 128 students, randomly distributed into a control group (n=64) and an experimental group (n=64). Execution time, compilable and executable tests, coverage of lines and branches, mutation score and detected defects were evaluated. Results: The experimental group reduced the mean resolution time from 44.8 to 36.2 minutes and obtained higher values of line coverage (82.6 % vs. 74.8 %), branch coverage (75.4 % vs. 66.1 %) and mutation score (70.8 % vs. 59.7 %). It also detected a greater number of defects, with statistically significant differences (p