Ricon Tsameret, Dolev-Cohen Michal
Faculty of Graduate Studies, Oranim Academic College of Education, 36006, Tivon, Israel.
Arch Sex Behav. 2025 Mar 11. doi: 10.1007/s10508-025-03100-x.
This study investigated the perspectives of artificial intelligence (AI) language models on the complex issue of sexting among adolescents, comparing their ability to mediate and manage quality communication on the topic with traditional parental mediation strategies. Drawing on a survey of 472 parents, the research compared parental attitudes, communication patterns, and mediation approaches to the outputs generated by four AI systems. The findings revealed that AI models emphasize to emphasize restrictive mediation strategies and highlight the perceived susceptibility and severity of sexting, while parental responses were varied. Notably, the AI systems differed in predicting dysfunctional parent-child communication on sexting, suggesting divergent implicit models of ideal communication dynamics. By comparing AI-generated responses with human reactions, this research offers initial insights into the significance of parental communication with adolescents regarding sexting and the potential limitations of current language models in promoting improved discussions between parents and adolescents on sensitive topics.
本研究调查了人工智能（AI）语言模型对青少年 sexting 这一复杂问题的看法，将它们就该主题进行调解和管理高质量沟通的能力与传统的父母调解策略进行了比较。基于对472名家长的调查，该研究将家长的态度、沟通模式和调解方法与四个人工智能系统生成的结果进行了比较。研究结果显示，人工智能模型强调限制性调解策略，并突出了 sexting 被感知到的易感性和严重性，而家长的反应则各不相同。值得注意的是，人工智能系统在预测关于 sexting 的功能失调的亲子沟通方面存在差异，这表明理想沟通动态的隐含模型存在分歧。通过将人工智能生成的反应与人类反应进行比较，本研究初步洞察了父母与青少年就 sexting 进行沟通的重要性，以及当前语言模型在促进父母与青少年就敏感话题进行更好讨论方面的潜在局限性。
