Value-Rich, Differently Organized: Measuring Value Expression and Organization in AI-Generated Debates
arXiv cs.AI · 21 Oct 2026, 03:48 AEST
Current evaluation of LLM-generated debate largely focuses on argumentative quality, leaving the value structure of generated arguments underexamined. This study introduces a diagnostic metric framework for comparing how values are expressed and organized in human-reference and AI-generated debates.
Key facts
- We introduce a diagnostic metric framework for comparing how values are expressed and organized in human-reference and AI-generated debates.
- The framework separates value expression, the range of recognizable values a debate invokes, from value organization, how those values are distributed and sustained across turns.
- Across multiple generation systems, AI-generated debates are broadly value-expressive, often activating as many or more value dimensions than matched human references, but they organize these values differently.
Read aloud
Current evaluation of L L M-generated debate largely focuses on argumentative quality, leaving the value structure of generated arguments underexamined. This study introduces a diagnostic metric framework for comparing how values are expressed and organized in human-reference and A I-generated debates. We introduce a diagnostic metric framework for comparing how values are expressed and organized in human-reference and A I-generated debates. The framework separates value expression, the range of recognizable values a debate invokes, from value organization, how those values are distributed and sustained across turns. Across multiple generation systems, A I-generated debates are broadly value-expressive, often activating as many or more value dimensions than matched human references, but they organize these values differently.
Sources in this cluster
- Value-Rich, Differently Organized: Measuring Value Expression and Organization in AI-Generated Debates
arXiv cs.AI · 21 Oct 2026, 03:48 AEST