New tool checks if AI chatbot test sets are actually good

Original: Benchmarking the Benchmarks: Evaluating Benchmarks for Conversational Agents

Source: arxiv.org ↗