Forum Working Groups
How AI is Reshaping Software Development & University Teaching
This group will explore how AI is reshaping software development, and how universities could adapt their teaching to better prepare graduates for industry. The goal is to gather the academic perspective on the same topic as the industry panel discussion. The combined academic and industry perspectives will then be synthesised into a white paper making recommendations to New Zealand universities.
Meaningful & Authentic Assessment for Students Learning to Program with AI
As we move into a new era of teaching programming and development skills to students which includes the use of AI tools, question arise about best practice for assessing these skills. This workshop aims to share current practice, future ideas and ongoing challenges in creating, running and evaluating both formative and summative assessments which include the use of AI.
Investigating source code analysis, generation, and repair
Contemporary approaches (e.g., LLMs) have advanced software engineering practices, particularly in source code analysis, generation, and repair. However, the performance of these approaches is still the topic of global debate. This working group will reflect on the effectiveness of these approaches, and identify avenues for improving the state of practice.
Automated Requirements & Acceptance Test Generation
This group will look into the current promises and pitfalls of Generative AI used to support requirements engineering activities — from refining user stories and eliciting user acceptance criteria, to generating automated test cases in a continuous integration loop. Further problems of interest include the sensitivity of model behaviour to prompt formulation and domain-specific vocabulary, the limited availability of benchmark datasets for requirements engineering tasks, and the challenge of evaluating generated artefacts against criteria such as completeness, testability, and freedom from ambiguity.
Testing and Engineering Agentic AI Software Systems
Agentic AI systems introduce new challenges for software development because they can plan, make decisions, use tools, interact with external systems, and adapt their behaviour at runtime. This working group will explore how agentic AI software can be designed, implemented, evaluated, and tested in reliable and trustworthy ways. Topics will include testing autonomous behaviours, validating agent workflows, managing non-determinism, assessing safety and correctness, monitoring deployed agents, and identifying the new engineering practices needed for developing AI agentic software systems.
