Survey reveals growing adoption of retrieval-augmented generation (RAG) to address AI risks
A new study on the state of large language models (LLMs) and retrieval-augmented generation (RAG) highlights the growing adoption of AI across organizations, despite concerns about data security, quality, and bias. The survey, sponsored by Graphwise, underscores the need for improved data governance and strategic AI implementation.
According to the State of Play on LLM and RAG: Preparing Your Knowledge Organization for Generative AI survey, 85% of organizations are either testing or actively deploying LLMs, with nine in ten planning to expand their implementations. However, 71% of respondents view the increased use of generative AI as a risk due to concerns over data security and output accuracy.
“AI’s potential is immense, but without rigorous data quality control, its benefits can be lost,”
Andreas Blumauer, Senior VP Growth at Graphwise.
To mitigate these risks, organizations are increasingly looking to RAG environments, which integrate structured and unstructured data for more precise AI-generated insights. Nearly one-third (29%) of surveyed companies are implementing RAG solutions to bridge the gap between their corporate databases and LLMs.
The survey, conducted by Unisphere Research in December 2024, included responses from 382 executives and managers overseeing knowledge management functions. Findings emphasize the strategic value of LLMs and RAG in enhancing decision-making, productivity, and customer experiences.
Graphwise, the result of a recent merger between Ontotext and Semantic Web Company, specializes in knowledge graph infrastructure and semantic AI technologies to support enterprise AI initiatives at scale.