Psychologically Potent, Computationally Invisible: LLMs Generate Social-Comparison-Eliciting Posts They Fail to Detect
arXiv:2605.01017v2 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 robustl
The proliferation of advanced LLMs necessitates research into their subtle yet impactful social interactions, particularly in generating content that elicits specific psychological responses.
This research highlights a significant limitation in LLM self-awareness and control over generated content's full psychological impact, crucial for ethical AI deployment and robust application development.
Our understanding of AI's ability to create socially potent content that the AI itself cannot robustly detect as such, indicating a gap between generation capabilities and analytical self-awareness.
- · AI ethics researchers
- · Social media platforms prioritizing well-being
- · Developers of advanced AI content moderation systems
- · LLM developers without robust psychological content detection
- · Users susceptible to social comparison content
- · Platforms reliant on basic sentiment analysis for content moderation
LLMs can generate content that subtly influences user psychology in ways they cannot detect or control.
This mismatch could lead to systems inadvertently amplifying harmful social comparison dynamics, necessitating new detection and control mechanisms.
Future AI development may focus more on self-reflective models capable of understanding the nuanced human psychological impact of their outputs.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL