This thesis investigates how tacit knowledge is discussed in academic discourse with aspecial focus on recent years and the introduction of generative AI. Tacit knowledge isunderstood here as knowledge that is neither formally documented nor codified inorganizational procedures. Instead, it emerges organically through practice, experience,and social interaction. Utilizing a dual-methodological framework that pairs qualitativethematic analysis, with computational natural language processing (corpus analysis), thisstudy maps a foundational corpus of 10,983 unique publications harvested from Scopus and Web of Science from 1968 to 2025. The longitudinal topography reveals a distinct W-shaped historical lifecycle. The literature transitioned from an initial cognitive and behavioral science root focused on implicit human learning (1968–1993), into a heavilycommercialized, extraction-oriented Knowledge Management (KM) paradigm focused oncorporate codification (1994–2010), before stabilizing into an enablement modelprioritizing social capital, psychological safety, and pedagogy (2011–2021).

To isolate the structural impact of recent technological breakthroughs, a specializedPRISMA-compliant sub-corpus of 30 core works and an expanded algorithmic database of521 entries specifically detailing the intersection of artificial intelligence and tacitknowledge were thoroughly analyzed. Pointwise Mutual Information (PMI) collocatemapping and transformer-based RoBERTa sentiment mining indicate a historic rebalancingof the field’s vocabulary. Rather than remaining an isolated peripheral node, AI hassolidified as a core thematic hub since late 2022. Contemporary scholarship exhibits astructural divergence, bypassing recent literature to link classic human-centric theoriesdirectly to modern automated pipelines. Sentiment profiling reveals a shift from defensiverejections of automation toward a cautiously optimistic, co-creative narrative. The findingsdemonstrate that while Large Language Models can instantly synthesize explicit data, theyremain fundamentally constrained by Polanyi’s paradox. Consequently, modern discoursereframes human intuition not as an automated target, but as an intractable, context-specificboundary and a mandatory check-and-balance required to ground black-box algorithmic recommendations.