Scholarlink

<p>This paper advances the notion of <i>coalescent agency</i> as a framework for understanding human–AI integration, thereby entering ongoing debates about machine agency, extended cognition, and AI governance. I argue that the persistence or erosion of human agency in human–AI systems can be predicted through four operational criteria constituting <i>normative directionality</i>: domain understanding, critical evaluation capacity, override authority, and responsibility attribution. Drawing on segmented ontology and predictive processing theory, I distinguish material-segment mechanisms (AI computational processing) from social-segment mechanisms (human normative practices) while showing how these heterogeneous structures can coordinate productively. The framework’s central prediction—that automation bias and agency erosion occur systematically when normative directionality criteria fall below measurable thresholds—connects philosophical analysis to empirical research and generates concrete design principles for agency-preserving AI systems.</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

The coalescent architecture of agency: normative directionality as the key to human–AI integration

  • 期刊论文

  • 发表日期：

    2026年3月19日

查看全文
  • Sergio Torres-Martínez

摘要

This paper advances the notion of coalescent agency as a framework for understanding human–AI integration, thereby entering ongoing debates about machine agency, extended cognition, and AI governance. I argue that the persistence or erosion of human agency in human–AI systems can be predicted through four operational criteria constituting normative directionality: domain understanding, critical evaluation capacity, override authority, and responsibility attribution. Drawing on segmented ontology and predictive processing theory, I distinguish material-segment mechanisms (AI computational processing) from social-segment mechanisms (human normative practices) while showing how these heterogeneous structures can coordinate productively. The framework’s central prediction—that automation bias and agency erosion occur systematically when normative directionality criteria fall below measurable thresholds—connects philosophical analysis to empirical research and generates concrete design principles for agency-preserving AI systems.