A Comprehensive Study on The Adoption of Large Language Models Across Industries

Abstract

Large language models (LLMs) have moved from experimental novelty to a standard component of enterprise technology stacks in under four years. This study synthesises survey data, industry reports, and peer-reviewed research to characterise how organisations across eight major sectors, healthcare, banking and financial services, legal services, retail and e-commerce, software development, customer service, education, and manufacturing, are deploying LLM-based tools, and with what measured effect. Drawing on McKinsey's 2025 global AI survey, the World Economic Forum's Future of Jobs Report, and sector-specific data from organisations including the International Legal Technology Association, GitHub, and peer-reviewed economics research, the study finds that adoption is now widespread, with 88% of organisations reporting regular use in at least one business function, but shallow: roughly two-thirds of adopters remain in pilot or experimentation phases, and few report measurable enterprise-level financial impact. Sector-level evidence shows the most rigorously documented productivity gains in customer service and software engineering, while legal, healthcare, and retail show rapid individual-level uptake that is outpacing organisational governance. The study concludes with an analysis of recurring barriers, including hallucination, data privacy, and regulatory uncertainty under frameworks such as the EU AI Act, and outlines priorities for closing the gap between adoption and measurable value.