FutureFinTech lecture: The art of generating useful tests

• Register
• Add to calendar

Description

Software testing is essential to building reliable, trustworthy systems, yet for most developers, working with automatically generated tests remains frustrating and time-consuming. In this FutureFinTech Lecture, Dr. Carolin Brandt presents research that puts developers back at the centre of the picture. 

Drawing on two empirical studies, Dr. Brandt examines what actually happens when developers encounter automatically generated tests in practice: what they accept, what they change, and why. The first study analyses feedback from 39 open-source project maintainers on amplified tests submitted for review. The second investigates tests generated by GitHub Copilot for Python projects, exploring how developers’ prompting choices affect the quality of output. 

For the financial services sector, where software reliability and compliance are non-negotiable, these findings carry direct relevance. Understanding how AI-assisted tools can be designed to work with developers rather than creating additional overhead is critical to realising their potential in high-stakes environments. 

Join us for expert insights and audience Q&A. This event will be in English.

About the speaker

Dr. Carolin Brandt, TU Delft

Dr. Carolin Brandt is an Assistant Professor in the Software Engineering Research Group (SERG) at TU Delft, where she leads the research line on Human- and Developer-centred Software Engineering. Her work focuses on building automation tools that genuinely serve the people who use them, examining not just whether tools perform well by technical measures, but whether developers can effectively work with what they produce. 

She completed her PhD cum laude from TU Delft in 2024 and is committed to fostering inclusive and constructive research communities. 

• About FutureFinTech