EOI Dataset — YouTube Discourse and Google Trends Corpus on Epistemic Outsourcing to Generative AI (2022–2026)

Epistemic Outsourcing Index (EOI) corpus measuring the balance between Epistemic Surrender (cognitive delegation to generative AI) and Epistemic Agency (critical engagement with AI output) in two complementary data streams. (1) YouTube SML corpus: 2,068 user comments collected April 2026 across 5 discourse frames (cognitive offloading, AI dependence, critical thinking loss, brain replacement, student AI use). (2) Google Trends longitudinal corpus: 4,683 weekly search-interest observations across 5 cognitive dimensions (Thinking, Writing, Memory, Critical Thinking, Creativity) and 5 anglophone markets (US, GB, AU, CA, IE) over 220 weeks (January 2022 to April 2026). Theoretical anchors: Clark and Chalmers (1998) Extended Mind Thesis; Sparrow, Liu, and Wegner (2011) Google Effect on Memory; Linden, Hachem, and Kondyli (2025) Homo Promptus framework; Wiederhold (2025) cognitive cost of AI convenience. Methodology: YouTube Data API v3 for SML; pytrends API for Google Trends. Mixed-methods design enabling triangulation between naturalistic discourse and search behaviour evidence.认知外包指数（Epistemic Outsourcing Index, EOI）语料库用于量化认知让渡（即向生成式AI（Generative AI）进行认知委托）与认知能动性（即对AI产出开展批判性互动）二者间的平衡，包含两个互补的数据流。(1) YouTube SML语料库：2026年4月采集的2068条用户评论，涵盖5类话语框架：认知卸载、AI依赖、批判性思维缺失、大脑替代、学生AI使用。(2) Google Trends纵向语料库：2022年1月至2026年4月（共计220周）的4683条周度搜索兴趣观测值，覆盖5个认知维度（思维、写作、记忆、批判性思维、创造力）以及5个英语国家市场（美国、英国、澳大利亚、加拿大、爱尔兰）。本研究的理论锚点包括：Clark与Chalmers（1998）提出的延展心智论题；Sparrow、Liu与Wegner（2011）提出的谷歌记忆效应；Linden、Hachem与Kondyli（2025）提出的“Homo Promptus”框架；Wiederhold（2025）提出的AI便捷性带来的认知成本。研究方法：采用YouTube Data API v3采集SML语料，使用pytrends API获取Google Trends数据。本研究采用混合研究设计，可实现自然主义话语与搜索行为证据之间的三角验证。