An Empirical Study on the Detection of Test Smells in Test Codes Generated by GitHub Copilot

Various techniques for automatically generating unit tests have been studied. The use of Large Language Models (LLMs) has recently emerged as a popular approach for automatic test generation from natural language descriptions. This study aims to measure the quality of the test codes produced by LLMs by detecting test smells in the test cases generate. To do this, we proposed an empirical study and a quality assessment methodology to be performed for each LLM that generates code. In our preliminary results, we applied these procedures with GitHub Copilot and obtained significant data on the quality of test codes. These findings indicate that although GitHub Copilot can generate valid unit tests, quality violations are still frequently found in these codes.