This study investigates how a large language model (LLM)-based chatbot can support novices in machine knitting, a domain that blends creative design, technical operation and digital programming. The research examines whether an LLM assistant can aid undergraduate fashion students with no prior knitting experience as they learn to use a digital knitting machine.
A custom chatbot was built using digital knitting machine manuals and integrated into an introductory machine-knitting course. Undergraduate students used the chatbot during hands-on training with the Kniterate machine. Data included chat logs, interviews with five students and an expert technical review. Analyses assessed the chatbot’s accuracy, usability and perceived educational value.
Students found the chatbot useful for basic troubleshooting, clarifying terminology and reinforcing classroom instruction. However, they preferred human guidance for complex, visual and judgment-based tasks. Expert evaluation revealed only 60% technical accuracy, and students questioned the chatbot’s reliability for nuanced machine-operation decisions.
The study offers an empirical examination of LLM assistance in a tactile, multimodal technical craft. It reveals both the emerging value and the current limitations of text-based AI for hands-on skill acquisition. By identifying gaps in visual reasoning and domain-specific precision, the research highlights opportunities for next-generation multimodal and specialized AI tools to better support technical learning environments.
