LLM Personification for Software Testing

Bachelor & Master Thesis

This research explores the use of Large Language Models (LLMs) to personify real human testers, enabling automated yet human-like software testing. Traditional automated testing often lacks the creativity, intuition, and contextual understanding that human testers bring. By leveraging LLMs’ advanced reasoning and natural language capabilities, this project aims to simulate diverse tester personas, ranging from novice users to expert developers, who can interact with software systems as real humans would. These AI-driven personas can identify usability issues, ambiguous requirements, and edge cases often missed by scripted tests. The study involves designing a framework where LLMs analyze user interfaces, generate test cases, and evaluate responses dynamically. Outcomes are expected to demonstrate improved test coverage, reduced manual effort, and more realistic user behavior simulation. This research bridges the gap between automated and human testing, contributing to smarter, adaptive, and context-aware quality assurance processes.

 

Required knowledge: python programming and data processing, experience with CV/NLP and (multimodal) LLMs, web crawling techniques, experience with mining software repositories, experience of automated GUI testing techniques.