Mapping the Landscape of Large Language Model (LLM) Adoption in Sub-Saharan Africa: A Systematic Cross-Database Bibliometric Study

Sub-Saharan Africa (SSA) is currently experiencing a technological compression in the domain of Large Language Models (LLMs), characterised by a hyper-accelerated scholarly response that masks deep structural fragilities. While the region is a prolific documenter of the socio-institutional impacts of generative AI, its visibility in the underlying computational engineering remains marginal, creating a dependency paradox. This study provides the first comprehensive macro-level diagnostic of the SSA LLM research landscape through a systematic cross-database bibliometric analysis. Utilising a PRISMA-compliant dataset of 255 high-quality documents retrieved from Scopus, Web of Science, and OpenAlex (2017-2026), the study employs science mapping and performance analysis to delineate the field's intellectual architecture. The data indicate a hyper-accelerated annual growth trajectory of 74.04% and a high international co-authorship rate (53.73%), yet expose a profound disciplinary asymmetry. While pedagogical sciences monopolise the intellectual core, technical engineering outputs suffer from near-zero citation impact. While recent cohorts show a natural citation lag, the data suggest the field is transitioning from an exploratory phase toward a descriptive plateau, necessitating a pivot toward indigenous technical engineering to sustain long-term impact. The study introduces the concept of reverse maturation to explain this reactive, event-driven evolution, arguing that the regional research agenda risks permanent theoretical marginalisation within the global knowledge economy. These results serve as critical evidence-based diagnostics for policymakers, highlighting the urgent necessity to pivot from reactive technology adoption toward indigenous computational architecture and sovereign innovation.