PresenceBench: Evaluating LLM Adaptation to User State
In brief
- PresenceBench measures LLM adaptation to user state (receptivity, flow, stability).
- Four models show human preference >0.5 for receptivity, but specificity remains unresolved.
- Using a response policy reduces costs and tokens without proven quality loss.
PresenceBench is a benchmark designed to assess how large language models adapt their responses based on the user's functional state (receptivity, flow, stability). The study evaluates receptivity and specificity when a state is provided, without measuring state detection.
This summary gives you the essentials. The full article, its figures and quotes are on cs.AI updates on arXiv.org.
Read the original articleAI-generated summary of the source, checked by rules. Report an error