Large Language Models in Multi Criteria Decision Making: A Systematic Review, Taxonomy, and Future Research Agenda

TL;DR

This is a systematic literature review that examines the integration of LLMs and MCDM, proposing a new taxonomy based on roles and integration stages.

Problem

There was a need to systematically organize and classify the emerging research landscape of LLM-MCDM integration, including its current status, trends, and remaining challenges.

Approach

Following PRISMA guidelines, 63 publications were selected from the Scopus database. Bibliometric analysis using VOSviewer and a literature review were conducted. A new classification system was developed based on the roles of LLMs, integration stages, and application areas.

Results & Contribution

The field has seen rapid growth since 2025, led by China and India. LLMs improve automation, scalability, and unstructured data processing in MCDM, but limitations such as bias, lack of interpretability, and prompt sensitivity were identified. The study concludes with research gaps and opportunities for advancing robust, explainable, and hybrid intelligent decision-making systems.