Orchestration logics for artificial intelligence platforms: From raw data to industry‐specific applications

Artificial intelligence (AI) platforms face distinct orchestra-tion challenges in industry-specific settings, such as theneed for specialised resources, data-sharing concerns, het-erogeneous users and context-sensitive applications. Thisstudy investigates how these platforms can effectivelyorchestrate autonomous actors in developing and consum-ing AI applications despite these challenges. Through ananalysis of five AI platforms for medical imaging, we identifyfour orchestration logics: platform resourcing, data-centriccollaboration, distributed refinement and application bro-kering. These logics illustrate how platform owners can ver-ticalize the AI development process by orchestrating actorswho co-create, share and refine data and AI models, ulti-mately producing industry-specific applications capable ofgeneralisation. Our findings extend research on platformorchestration logics and change our perspective fromboundary resources to a process of boundary processing.These insights provide a theoretical foundation and practi-cal strategies to build effective industry-specific AIplatforms.     «

Artificial intelligence (AI) platforms face distinct orchestra-tion challenges in industry-specific settings, such as theneed for specialised resources, data-sharing concerns, het-erogeneous users and context-sensitive applications. Thisstudy investigates how these platforms can effectivelyorchestrate autonomous actors in developing and consum-ing AI applications despite these challenges. Through ananalysis of five AI platforms for medical imaging, we identifyfour orchestration logics: platform...     »