Workshop Program

09:00 - 09:10	Workshop Opening
09:10 - 09:40	Keynote The Real Impact of AI Tools on Software Development: An Industry Perspective Steve Haupt, Andrena AI tooling is transforming software engineering, offering unprecedented opportunities for innovation and efficiency, but it also brings significant risks. This talk provides an industry perspective on the spectrum of AI tools in use today, showcasing what is possible when cutting-edge tools are embraced to their full potential, and the measurable efficiency gains they deliver.
09:40 - 10:10	Presentation of accepted contribution Code Generation for Niche Programming Languages with Large Language Models Philipp Kogler, Wei Chen and Stefan Wallner (Siemens Österreich)
10:10 - 10:40	Presentation of accepted contribution Stories Complicate Things: A Qualitative Analysis of Coding Problems (Un)solved by GitHub Copilot Julian Oertel (University of Rostock), Jil Klünder (Leibniz University Hannover) and Regina Hebig (University of Rostock)
10:40 - 11:10	Coffee Break

11:10 - 11:40	Keynote AI-Assisted Correctness-by-Construction Software Development Maximilian Kodetzki, Karlsruhe Institute of Technology In recent years, research on artificial intelligence has made great progress. AI-tools are getting better in simulating human reasoning and behavior every day. We discuss the extent to which AI-tools can support Correctness-by-Construction (CbC) engineering. This is an approach of formal methods for developing functionally correct programs incrementally on the basis of a formal specification. Using sound refinement rules, the correctness of the constructed program can already be guaranteed in the development process. We analyze the CbC development process regarding steps for potential AI-tool support in the tool CorC, which implements CbC. We classify the findings in five areas of interest: refinement process, specification generation, code generation, deductive verification, and interpretation. Based on existing work, expert knowledge, and prototypical experiments, we discuss for each of the areas whether and to what extent AI-tools can support CbC software development. We address the risk of AI-tools in formal methods and present our vision of AI-integration in the tool CorC to support developers in constructing programs using CbC engineering.
11:40 - 12:10	Presentation of accepted contribution On the logical (in)consistency of code-generating LLMs Ke Dong, William Hsu, Pascal Hitzler and Eugene Vasserman (Kansas State University, USA)
12:10 - 12:30	Reflexion round & end of workshop