Midi de la recherche SIO avec Taha Havakhor

À propos de
l'événement

Le Département de systèmes d’information organisationnels vous invite à une présentation du professeur Taha Havakhor sur sa recherche portant sur la création de valeur organisationnelle via l’IA générative.

La présentation se déroulera en présentiel et en anglais.

Une boîte à lunch sera offerte gratuitement aux personnes inscrites.

Résumé

Why Do Firms Realize Different Returns from Public Generative AI? A Two-Channel Theory of Organizational Value Creation

Generative artificial intelligence (GenAI) combines two characteristics that distinguish it from previous generations of information technology: foundation models are both ubiquitously accessible and capable of progressively incorporating firm-specific organizational knowledge. This combination fundamentally changes how organizations create value from digital technologies. Existing business value of information technology (BVIT) theories explain value creation through complementary organizational resources surrounding relatively stable technologies, but they cannot fully explain GenAI because organizations can increasingly embed proprietary knowledge, governance mechanisms, and organizational learning within the technology itself. We develop a dual-channel theory of GenAI value creation. The passive channel captures productivity gains arising from employees’ exposure to GenAI-compatible tasks through the use of public foundation models, while the active channel captures organizational investments that progressively transform those public models into firm-specific organizational capabilities. Within the active channel, we distinguish between integration, which improves the use of public foundation models, and transplantation, whereby organizations internalize proprietary knowledge through retrieval infrastructures, fine-tuning, governance mechanisms, and continuous organizational learning. We evaluate this framework using complementary quantitative and qualitative evidence. Using archival data on publicly traded firms, occupational exposure measures, and GenAI-related job postings, we show that occupational exposure generates significant productivity gains, particularly among firms possessing stronger IT and AI capabilities, and induces organizations to reconfigure their workforce toward AI- and IT-intensive occupations, consistent with capability development. We further develop a novel four-layer measure of organizational GenAI investment and demonstrate that productivity gains are substantially greater for transplantation-oriented investments involving retrieval-augmented generation and model adaptation than for integration-oriented investments such as generic usage and prompt engineering. Qualitative case evidence illustrates how organizations construct governance structures, knowledge infrastructures, and feedback mechanisms that progressively embed proprietary organizational knowledge into AI capabilities. By distinguishing productivity opportunities created through ubiquitous public AI from firm-specific value created through transplantation, this study extends the BVIT literature by showing that GenAI introduces a new strategic decision space in which organizations must choose not only whether to invest in AI, but how to transform a universally accessible technology into a proprietary organizational capability.