Can I Teach Graph Rewriting to My Chatbot?

Chatbots based on Large Language Models have been making progress not just in traditional natural language-processing tasks but also in generating from informal requirements formal artefacts such as models and programs. In this short paper we report on experiments in the use of GPT-4 for creating graph rewriting systems. Through two case studies, we guide the bot in the design, extension and analysis of type graphs and rewrite rules in the GXL format of the Groove graph rewriting tool, which we use to visualise our models. While the current experiments are entirely manual, they provide anecdotal evidence for the potential of using large language models in creating and analysing graph rewriting systems and show how a simple tool integration could support a natural language interface for graph rewriting tools.