Large Language Model Adoption: Systematic Review, Theoretical Frameworks, and Meta-Analytic Evidence

The adoption of large language models (LLMs) is reshaping how organizations approach automation, decision-making, and user engagement across sectors. This study investigates the trends, theoretical frameworks, and adoption factors influencing the integration of LLMs in five key domains: education, commerce, banking, healthcare, and service. By employing a systematic literature review and meta-analysis, this paper synthesizes research published between 2022 and early 2026, corresponding to the period when LLMs became widely accessible for public and enterprise use, to evaluate both conceptual and empirical dimensions of LLM adoption. The review identifies the Technology Acceptance Model and the Unified Theory of Acceptance and Use of Technology, including its extensions, as the most frequently applied frameworks. It also highlights the growing incorporation of complementary models such as the diffusion of innovation, the information system success model, and self-determination theory. The meta-analysis examines 59 pairwise relationships drawn from 154 studies with a cumulative sample size of 88,886 participants. Using correlation coefficients, I2 statistics, and Egger’s test, the analysis reveals strong, consistent associations between behavioral intention and both use behavior and actual use, while also identifying high heterogeneity across contexts. Constructs such as trust, hedonic motivation, and personal innovativeness emerged as influential but were underrepresented in the theoretical modeling. The study underscores the importance of facilitating conditions, infrastructure, and organizational readiness for enabling sustained use while also drawing attention to gaps in addressing perceived risks, privacy concerns, and ethical implications.