Knowledge management has cycled through several technological waves—databases, intranets, enterprise social networks—each promising to solve the perennial problem of making organizational knowledge accessible to those who need it. Generative AI represents the most disruptive wave yet, because it does not merely store or retrieve knowledge but synthesizes, translates, and generates new knowledge artifacts from existing organizational data.
Zhang, Zuo, and Yang (2025) investigate GenAI's impact on enterprise innovation performance from a knowledge management perspective, developing a theoretical framework that calls for further empirical validation. Their findings reveal a dual mechanism: GenAI enhances innovation by accelerating knowledge combination (connecting previously unrelated knowledge domains) while simultaneously creating new challenges around information overload, data bias amplification, and technological dependency. The innovation gains are real but not automatic—they require what the authors call "absorptive capacity for AI output," meaning the organizational ability to evaluate, contextualize, and critically assess AI-generated knowledge rather than accepting it uncritically. Firms that deployed GenAI without this absorptive capacity experienced what might be called "confident ignorance": AI-generated analyses that were plausible, fluent, and wrong, adopted precisely because they sounded authoritative.
He and Burger-Helmchen (2025) take a more theoretical approach, examining how AI reshapes the foundational knowledge management frameworks, particularly the SECI model that has guided KM practice for three decades. Their analysis suggests that AI is strongest at supporting externalization (converting tacit knowledge to explicit through automated documentation, transcription, and pattern extraction) and combination (merging explicit knowledge sources into new syntheses). However, AI is weakest at socialization (the tacit-to-tacit knowledge transfer that occurs through shared experience and mentorship) and internalization (the process by which individuals absorb explicit knowledge into personal practice). This asymmetry matters because the most strategically valuable organizational knowledge—judgment, intuition, relational awareness—resides precisely in the dimensions where AI's contribution is most limited.
He, Yousaf, and Palazzo (2025) examine the synergistic effects of organizational innovativeness, knowledge sharing, AI adoption, and big data analytic capabilities within human resource management. Their structural equation modeling reveals that AI adoption amplifies the benefits of existing knowledge-sharing cultures but does not substitute for them. Organizations with weak knowledge-sharing norms saw limited innovation gains from AI adoption because the AI had less organizational knowledge to work with—the garbage-in, garbage-out principle applied at the organizational level. Conversely, organizations with strong knowledge-sharing cultures but limited AI adoption were leaving accessible performance gains on the table.
The synthesis points toward a "both-and" imperative. GenAI can dramatically accelerate knowledge management processes, but only in organizations that have already built the social and cultural infrastructure for knowledge sharing. Technology and culture are complements, not substitutes. The organizations most likely to realize AI's knowledge management potential are those that invest simultaneously in AI tools and in the human practices—trust, psychological safety, incentives for sharing—that generate the knowledge AI needs to be useful.