AI in Service of Software Quality: How ChatGPT and Personas Are Transforming Exploratory Testing

• Ágatha de Almeida UFAM
• Eliane Collins INDT
• Ana Carolina Oran UFAM

Resumo

Context: Exploratory testing is essential in the software validation process as a way to find unexpected and critical failures in a short time, complementing documented functional test cases. However, creating scenarios to explore the software (such as test charters) can be time-consuming, and depending on the team’s experience, it may lack adequate coverage of functionalities and scenarios that target specific user profiles of the application. Objective: This article investigates how AI, through LLMs (Large Language Models), can assist in creating exploratory test charters that reflect the characteristics and needs of different user personas. Method: To achieve this, an experimental study was conducted where personas were used as input in ChatGPT 3.5 to generate exploratory test charters. The effectiveness of the approach was evaluated by Software Engineering students, who analyzed the performance and usefulness of the generated charters through a questionnaire based on the TAM model, supplemented by qualitative and quantitative analyses. Results: Data analysis indicated positive acceptance of ChatGPT 3.5 by the participants, highlighting its ease of use and perceived usefulness. Conclusion: This study contributes to the field of Software Engineering by demonstrating a practical application of artificial intelligence in the automated generation of test charters. ChatGPT 3.5 has proven to be a promising tool to support the creation of personalized exploratory test charters, contributing to software quality improvement. The integration of artificial intelligence techniques with user-centered design methods can significantly optimize the software testing process.

Palavras-chave: Exploratory Testing, ChatGPT, Personas, Software Quality, Artificial Intelligence