CompanionBench, a real-data-grounded bilingual benchmark with a hidden disclosure gate and IRT-corrected judging, ranks 28 AI companions and finds most fail to earn deeper disclosure, often substituting warmth for substance.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CompanionBench: A Theory-Anchored, Real-World-Grounded Benchmark for AI Emotional Companionship
CompanionBench, a real-data-grounded bilingual benchmark with a hidden disclosure gate and IRT-corrected judging, ranks 28 AI companions and finds most fail to earn deeper disclosure, often substituting warmth for substance.