PI: Per Runeson, Lund University
co-PI: Christian Kowalkowski, Linköping University
With the increasing use of AI models in society and business, the issue of using general vs. specialized models (e.g. per task, company, country or domain) is raised. From a business perspective, it is a matter of control over the supply chain for service provisioning, and the ability to protect the basis for differentiation of the market offering. From a technical perspective, there are a multitude of factors to trade across the range of generalization–specialization, including model performance, energy consumption, training data supply, and integrity of data. Against this backdrop, this dual-legged project aims to address innovation and engineering for AI-enabled services in a B2B context. Both legs use empirical studies of industry practice as a source of in-depth problem understanding. The innovation leg aims to provide recommendations for business practice on strategies for service innovation in different ecosystems contexts, and the engineering leg aims to design solutions for technical special- ization and ecosystem governance. The SPAISIE project builds on existing collaboration between LU and LiU in a previous ELLIIT project, shifting focus from data sharing to addressing AI models in complex business-to-business (B2B) settings. It aligns particularly well with the ELLIIT technology foresight areas of digital business models, and DevOps and MLOps.
Project number: F12