How should AI based automated test generation make us rethink the software testing pyramid ? We compare two automated test generation approaches (LLM vs SBST), discuss open source tools & industry use cases, and provide developers with recommendations for integrating AI into their testing workflows.

Cohn’s Test Pyramid has been a foundational aspect of agile development, guiding developers in the delicate balance between unit, integration, and e2e testing. The recent surge in the adoption of AI based code generation necessitates a need to rethink the canonical form of the testing pyramid as test driven development takes on a “shift-left” approach, pushing complex integration testing to earlier in the workflow. In this talk, we start by discussing the two main automated test generation approaches in use today (LLM-based vs SBST-based) and tradeoffs between them, and then we dive into some open source generation software and current industry use-cases using AI based test generation. Finally, we address some common myths surrounding future speculation about advancements in this area, and close with informed recommendations for developers on when to consider AI based solutions for your testing workflows and which approach to use. Key takeaways for developers: - A practical understanding of the tradeoffs between LLM and SBST-based test generation approaches. - Available test generation tools that can be used for your use-cases. - Recommendations for when and how to integrate automated test generation in your workflow

Hi, I’m Surbhi, currently working as a Senior Software Engineer at Google in NYC. My background includes 7 years of experience working in Android native client development, platform engineering, performance optimization, backend architectures, and server-driven UIs all of which have involved test-driven development. My work has primarily been in Java and C++. I graduated from Brown University with a degree in Computer Science where I contributed to the department’s undergraduate teaching in data structures algorithms, focusing on making courses inclusive and collaborative and introducing unit testing and the testing pyramid paradigm to the course curriculum, which has since become a mainstay fixture. Since then, I have at Google, contributed extensively towards intern hiring, mentoring early career professionals, and helping foster inclusive team cultures. I enjoy following pro tennis, running, biking, cooking, board games, walking, eating good food, and exploring the city. Reach me at surbhimadan1995@gmail.com .