Scholarlink

<p>Artificial intelligence (AI) is rapidly transforming the scientific enterprise, yet its effects on knowledge dynamics—innovation and diffusion—remain poorly understood in the social science domain. Focusing on public administration scholarship, this study assesses whether AI integration is associated with changes in innovation and diffusion within the field. Using public administration articles indexed in Web of Science (1923–2025), we identify AI-integrated studies and construct matched comparison sets of non-AI articles with similar publication characteristics. We measure innovation using lexical and semantic novelty derived from bibliometric metadata and large language model–based semantic representations, and measure diffusion using citation-based indicators of academic influence and diffusion speed. AI-integrated articles exhibit significantly higher novelty, greater citation impact, and faster diffusion than matched non-AI studies. Decomposition analyses suggest that these gains are driven primarily by combinatorial innovation rather than the introduction of entirely new foundational ideas. Semantic trajectory analyses further show that AI-integrated work diverges more from established research pathways, consistent with reduced path dependence. This study extends technology-induced innovation debates into the social sciences, and offers scalable methodological tools for tracing AI’s imprint on scholarly communication.</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

Innovation and Diffusion: The Effects of AI Integration on Knowledge Dynamics in Public Administration Scholarship

  • 期刊论文

  • 发表日期：

    2026年3月31日

查看全文
  • Dehu Yin,
  • Yu Wang,
  • Hongxu Liu,
  • Zhaopeng Li,
  • Li Tang

摘要

Artificial intelligence (AI) is rapidly transforming the scientific enterprise, yet its effects on knowledge dynamics—innovation and diffusion—remain poorly understood in the social science domain. Focusing on public administration scholarship, this study assesses whether AI integration is associated with changes in innovation and diffusion within the field. Using public administration articles indexed in Web of Science (1923–2025), we identify AI-integrated studies and construct matched comparison sets of non-AI articles with similar publication characteristics. We measure innovation using lexical and semantic novelty derived from bibliometric metadata and large language model–based semantic representations, and measure diffusion using citation-based indicators of academic influence and diffusion speed. AI-integrated articles exhibit significantly higher novelty, greater citation impact, and faster diffusion than matched non-AI studies. Decomposition analyses suggest that these gains are driven primarily by combinatorial innovation rather than the introduction of entirely new foundational ideas. Semantic trajectory analyses further show that AI-integrated work diverges more from established research pathways, consistent with reduced path dependence. This study extends technology-induced innovation debates into the social sciences, and offers scalable methodological tools for tracing AI’s imprint on scholarly communication.