analiza aiqbench
This research reveals a critical blind spot in large language models where they can generate socially potent content that simultaneously evades their own detection mechanisms, highlighting a fundamental gap between generation and understanding. The study establishes a new benchmark showing that LLMs struggle to robustly classify subtle social comparison cues embedded in text, suggesting current models may inadvertently amplify relational dynamics without recognizing them. This finding is vital for safety researchers concerned with how AI systems might manipulate or misinterpret human social signals in real-world applications.
Psychologically Potent, Computationally Invisible: LLMs Generate Social-Comparison-Eliciting Posts They Fail to Detect
arXiv:2605.01017v3 Announce Type: replace Abstract: We introduce Xiaohongshu Social Comparison Reader Elicitation (XHS-SCoRE), a reader-grounded benchmark for detecting whether text-only Xiaohongshu (RedNote) posts elicit Upward, Downward, or Neutral/no clear social comparison from a first-person reader perspective. The task targets a socially meaningful relational, behaviorally real signal not reducible to sentiment. Across prompted LLM classifiers and supervised Chinese encoders, we find a consistent generation-detection mismatch: the signal is textually learnable in-domain, but not robustly accessible to prompt-based classification. Prompted LLM classifiers show stable failures, especially neutralization of comparison-eliciting posts and model-specific directional skew. A controlled pilot shows that LLM-generated Xiaohongshu-style posts can shift perceived standing and comparison-related affect even when prompt-based detection of the same construct remains fragile. XHS-SCoRE contributes a benchmark for reader-grounded comparison detection and a diagnostic framework for studying when socially meaningful relational cues remain only partially visible to prompt-based inference.
Abstract:We introduce Xiaohongshu Social Comparison Reader Elicitation (XHS-SCoRE), a reader-grounded benchmark for detecting whether text-only Xiaohongshu (RedNote) posts elicit Upward, Downward, or Neutral/no clear social comparison from a first-person reader perspective. The task targets a socially meaningful relational, behaviorally real signal not reducible to sentiment. Across prompted LLM classifiers and supervised Chinese encoders, we find a consistent generation-detection mismatch: the signal is textually learnable in-domain, but not robustly accessible to prompt-based classification. Prompted LLM classifiers show stable failures, especially neutralization of comparison-eliciting posts and model-specific directional skew. A controlled pilot shows that LLM-generated Xiaohongshu-style posts can shift perceived standing and comparison-related affect even when prompt-based detection of the same construct remains fragile. XHS-SCoRE contributes a benchmark for reader-grounded comparison detection and a diagnostic framework for studying when socially meaningful relational cues remain only partially visible to prompt-based inference.
Accepted at the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026). 20 pages, including references and appendices; 1 figure and 6 tables. Project repository: this https URL
Computation and Language (cs.CL)
arXiv:2605.01017 [cs.CL]
(or arXiv:2605.01017v3 [cs.CL] for this version)
Text integral sub licența deschisă a publicației · 180 cuvinte
Subiecte conexe
Game Design Creativity evals
LessWrong · 12 oct. 2026
Links #3: 2026/06 Part 1
LessWrong · 15 iun. 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · 13 oct. 2026
Reviewing failure modes in science as we prepare for AI researchers
LessWrong · 8 oct. 2026