Scholarlink

<p>The rapid expansion of large language models (LLMs) has intensified debates regarding their use, their psycho-emotional implications, and the underlying risks they entail. Although recent scholarship has examined attitudes toward artificial intelligence and documented the emergence of affective uses, little is known about the relationship between declared attitudes and actual interactional practices with LLMs. This article advances a conceptual and empirical model of symmetry and asymmetry to explain why favorable or unfavorable positions toward the emotional uses of generative artificial intelligence (GenAI) do not necessarily predict the breadth or intensity of affective engagement. Drawing on a concurrent mixed methods design involving 285 survey responses and 35 semi-structured interviews with Mexican university students aged 18–24, this study combines the results of an attitudinal scale (EAUE-GenAI; <i>α</i> = 0.90) with a discourse analysis of interactional repertoires. Quantitative findings indicate low-to-moderate, subtly conservative affective attitudes (<i>M</i> = 2.47/5), whereas qualitative analysis identifies a progressive scale of emotional engagement: (1) emergent emotional advice, (2) validation, and (3) anthropomorphization. These findings demonstrate that affective practices frequently exceed declared instrumental positions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Symmetries and asymmetries between attitudes and interaction in relation to the emotional uses of LLMs

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月11日

查看全文
  • Juan Pablo Duque Parra,
  • Alejandro Santes Ortega

摘要

The rapid expansion of large language models (LLMs) has intensified debates regarding their use, their psycho-emotional implications, and the underlying risks they entail. Although recent scholarship has examined attitudes toward artificial intelligence and documented the emergence of affective uses, little is known about the relationship between declared attitudes and actual interactional practices with LLMs. This article advances a conceptual and empirical model of symmetry and asymmetry to explain why favorable or unfavorable positions toward the emotional uses of generative artificial intelligence (GenAI) do not necessarily predict the breadth or intensity of affective engagement. Drawing on a concurrent mixed methods design involving 285 survey responses and 35 semi-structured interviews with Mexican university students aged 18–24, this study combines the results of an attitudinal scale (EAUE-GenAI; α = 0.90) with a discourse analysis of interactional repertoires. Quantitative findings indicate low-to-moderate, subtly conservative affective attitudes (M = 2.47/5), whereas qualitative analysis identifies a progressive scale of emotional engagement: (1) emergent emotional advice, (2) validation, and (3) anthropomorphization. These findings demonstrate that affective practices frequently exceed declared instrumental positions.