Explainable Software Engineering: Rethinking the full SE life cycle

This will be a dry run of my keynote “Explainable Software Engineering” at the Bits & Chips Event, which will take place next week in Eindhoven.

Abstract

In artificial intelligence (AI), it’s increasingly recognized that components that learn from data need to be explainable. In this talk, we take explainability one step further, using it as a lens to rethink the full software engineering life cycle. To that end, we consider explainability of both the software engineering process and the resulting software system. We use this to shed new light on requirements traceability, delay prediction, code review and AI-powered coding. Furthermore, we revisit software testing, interpreting test cases as executable explanations at different levels of granularity. Based on this, we envision a future of software engineering in which explainability is a first-class citizen.

Bio

Arie van Deursen is a professor in software engineering at Delft University of Technology, where he’s also head of the Software Technology department. He holds an MSc degree from the Vrije Universiteit Amsterdam (1990) and a PhD from the University of Amsterdam (1994). His research interests include software testing, language models for code, human aspects of software engineering and trustworthy artificial intelligence. He’s scientific director of the Delft Fintech Lab and co-PI of the NWO Long Term Program Robust (2022-2032) on trustworthy AI. Based on his research, he co-founded the Software Improvement Group (2000) and PerfectXL (2014). For the Dutch government, he serves on the Advisory Council for ICT Assessments (AcICT).