Annotation alignment: Comparing LLM and human annotations of conversational safety

Open in new window