Study finds simpler AI explanations help humans verify outputs better than preferred formats

Original: Do Reasoning Representations Help Humans Evaluate LLM Outputs?

Source: arxiv.org ↗