Consistency in Language Models: Current Landscape, Challenges, and Future Directions
It addresses the challenge of unreliable consistency in language models for AI applications, but is incremental as it reviews existing work rather than proposing new solutions.
The paper examines the problem of language models struggling to maintain consistency across tasks and domains, identifying research gaps and calling for better benchmarks and interdisciplinary approaches.
The hallmark of effective language use lies in consistency: expressing similar meanings in similar contexts and avoiding contradictions. While human communication naturally demonstrates this principle, state-of-the-art language models (LMs) struggle to maintain reliable consistency across task- and domain-specific applications. Here we examine the landscape of consistency research in LMs, analyze current approaches to measure aspects of consistency, and identify critical research gaps. Our findings point to an urgent need for quality benchmarks to measure and interdisciplinary approaches to ensure consistency while preserving utility.