Governing generative AI for tacit knowledge retention: a problem theory

Generative AI (GenAI) promises to capture and transfer tacit knowledge at scale through interactive dialogue and personalised learning support. However, GenAI's architectural properties introduce novel challenges that demand re-problematisation of knowledge retention. We contribute a Problem Theory that reframes knowledge retention from a paradigm selection challenge into a governance design challenge. While traditional strategies force trade-offs between personalisation's depth and codification's scale, GenAI as a conduit between experts and novices transforms the shared context enabling knowledge transfer, introducing distinct epistemic, motivational, and pedagogical risks. We argue that achieving symbiosis between GenAI capabilities and human expertise requires governance mechanisms addressing both traditional retention challenges and these novel GenAI-specific risks. Through systematic decomposition across three stakeholder relationships, we identify the needs, goals, and governance requirements for tacit knowledge transfer at scale. This problematisation contributes to the emerging GenAI-mediated knowledge retention domain as the conceptual foundation for future empirical validation and design work.