Scholarlink

<p>The rapid expansion of generative artificial intelligence (GenAI) systems has intensified debates about their role in knowledge production and education. While much of the existing research focuses on performance metrics or algorithmic bias, less attention has been paid to how these systems are negotiated in intercultural contexts where linguistic diversity and community-based knowledge shape epistemic legitimacy. This article examines the interaction between Tseltal students and large language models during a two-day literary creation workshop conducted at the Universidad Intercultural de Chiapas in Oxchuc, Mexico. The study employs the Operative Research Group methodology to observe how participants collectively engage with generative outputs through narrative decomposition, iterative prompting, bilingual experimentation, and collaborative evaluation. The findings show that when prompts lacked contextual specificity, generated narratives tended to reproduce globally dominant story templates. When interaction occurred in Tseltal, infrastructural asymmetries became visible through translation-first responses and grammatical inconsistencies. Participants actively corrected these outputs by reintroducing locally grounded knowledge, territorial references, and cultural procedures. The study argues that the key issue in intercultural uses of GenAI is not simply the presence of algorithmic bias but the negotiation of epistemic authority between probabilistic outputs and culturally situated human judgment. The findings highlight the importance of human agency as a mediating condition for maintaining epistemic plurality in contexts of linguistic and cultural diversity.</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

Who speaks for culture? Generative AI, epistemic authority, and narrative negotiation in an intercultural classroom

  • 期刊论文

  • 发表日期：

    2026年5月30日

查看全文
  • João Gabriel Rodrigues Telles Almeida

摘要

The rapid expansion of generative artificial intelligence (GenAI) systems has intensified debates about their role in knowledge production and education. While much of the existing research focuses on performance metrics or algorithmic bias, less attention has been paid to how these systems are negotiated in intercultural contexts where linguistic diversity and community-based knowledge shape epistemic legitimacy. This article examines the interaction between Tseltal students and large language models during a two-day literary creation workshop conducted at the Universidad Intercultural de Chiapas in Oxchuc, Mexico. The study employs the Operative Research Group methodology to observe how participants collectively engage with generative outputs through narrative decomposition, iterative prompting, bilingual experimentation, and collaborative evaluation. The findings show that when prompts lacked contextual specificity, generated narratives tended to reproduce globally dominant story templates. When interaction occurred in Tseltal, infrastructural asymmetries became visible through translation-first responses and grammatical inconsistencies. Participants actively corrected these outputs by reintroducing locally grounded knowledge, territorial references, and cultural procedures. The study argues that the key issue in intercultural uses of GenAI is not simply the presence of algorithmic bias but the negotiation of epistemic authority between probabilistic outputs and culturally situated human judgment. The findings highlight the importance of human agency as a mediating condition for maintaining epistemic plurality in contexts of linguistic and cultural diversity.