Structural centrality of artificial intelligence and technology transfer: evidence from pharmaceutical innovation labs
研究了大型制药公司如何通过创新实验室组织人工智能驱动的知识和技术转移，发现AI的结构中心性影响治理模式、合作方式和专有性条件。
Abstract Artificial intelligence is increasingly reshaping pharmaceutical research and development, yet limited attention has been paid to how firms organise the integration and governance of AI-related capabilities. This study examines how large pharmaceutical companies structure innovation labs as organisational interfaces for mobilising and coordinating AI-driven knowledge and technologies. Building on technology transfer and organisational design theory, we conceptualise the structural centrality of artificial intelligence within innovation labs as a key design variable shaping governance arrangements, collaboration patterns, and appropriability conditions. Drawing on an original dataset of 138 innovation labs established by 24 of the world’s largest pharmaceutical companies, we classify labs according to whether artificial intelligence is absent, embedded within broader innovation portfolios, or structurally centralised in dedicated units. Using a configuration-based descriptive approach, we analyse differences in strategic orientation, collaborative models, and activity profiles across these configurations. The findings are consistent with the view that variation in structural centrality is associated with systematically different patterns of capability governance. Labs in which artificial intelligence is embedded appear to rely on diffusion-oriented and partnership-based arrangements, supporting cross-domain experimentation. In contrast, dedicated artificial intelligence labs display characteristics consistent with more internalised and governance-intensive configurations. Moreover, AI-focused infrastructures are highly concentrated among the largest firms, suggesting scale-dependent conditions for orchestrating data-intensive capabilities. By conceptualising innovation labs as differentiated organisational interfaces for AI-related technology transfer, this study extends contingent models of technology transfer into the domain of digital transformation in regulated, science-based industries.