Extending hybrid intelligence theory: dynamic boundary negotiation in professional human-AI collaboration within design practice

Generative artificial intelligence (GenAI) is reshaping professional design practice; however, existing studies have focused on static human-AI complementarity with limited attention to how professionals maintain agency during dynamic collaboration. This longitudinal study tracked three senior design professionals collaborating with GenAI over six months and analyzed work diaries and interviews through an abductive logic informed by constructivist grounded theory. The findings revealed a three-layer boundary structure (Capability, Role, and Control Boundaries) through which professionals maintain their agency. These boundaries undergo dynamic restructuring in response to task changes and technological changes, with differentiated adaptation strategies emerging according to disruption sources. This study proposes the DANCE (Dynamic Agency Negotiation in Collaborative Expertise) model, reconceptualizing human-AI collaboration as ongoing boundary negotiation between professional and algorithmic agency. An ‘Efficiency Paradox’ was identified, in which professionals deepened their engagement with AI collaboration despite persistent operational difficulties. This pattern suggests that the negotiation process itself yields returns related to professional identity and value positioning beyond task efficiency. Cross-case analysis indicated that professional value shifted from the ‘power to execute’ toward the ‘power to architect’. This study extends hybrid intelligence and boundary work theories, offering implications for design education and organisational practice.