The launch of ChatGPT by OpenAI has caused waves across industries, revolutionising the way we interact with artificial intelligence and sparking discussions on the future of technology, ethics, and human-machine collaboration. Accounting educators must find innovative ways to integrate artificial intelligence text generators (AITG) into the accounting curricula and into their teaching practices. Large Language Models, such as ChatGPT, have the potential to support student understanding of complex disciplinary concepts in an interactive and engaging way. In this study, we contribute to the rapidly growing literature on the use of AITG in accounting education. Using threshold concepts in the specialised field of Business Combinations and Consolidations (BCC), we explore whether ChatGPT is a reliable source for knowledge creation (knowing that and knowing how) in accounting. When prompted to explain different threshold concepts in BCC, the AITG generated responses which demonstrated the ability to provide epistemic knowledge which may aid disciplinary understanding (knowing that); however, when prompted to apply procedural knowledge (knowing how), the generated responses were superficial and mainly incorrect. Our study finds that an AITG can be a useful learning resource when used in a controlled environment, but on its own, it is not sufficient to support student mastery of threshold concepts.