Understanding The Effect Of Temperature On Alignment With Human Opinions
This addresses the problem of aligning LLM outputs with human opinions for researchers and practitioners, but it is incremental as it builds on existing methods.
The study investigated how different methods for extracting opinion distributions from LLMs align with human opinions, finding that sampling and log-probability approaches with parameter adjustments yield better alignment in subjective tasks compared to direct prompting.
With the increasing capabilities of LLMs, recent studies focus on understanding whose opinions are represented by them and how to effectively extract aligned opinion distributions. We conducted an empirical analysis of three straightforward methods for obtaining distributions and evaluated the results across a variety of metrics. Our findings suggest that sampling and log-probability approaches with simple parameter adjustments can return better aligned outputs in subjective tasks compared to direct prompting. Yet, assuming models reflect human opinions may be limiting, highlighting the need for further research on how human subjectivity affects model uncertainty.