The use of LLM assistants for kernel development