Exploring the Potential of Large Language Models in Self-adaptive Systems

TL;DR

The paper addresses whether large language models can enhance self-adaptive systems by conducting a cross-disciplinary literature review across software engineering, autonomous agents, robotics, and autonomous driving. It classifies findings using the MAPE-K framework plus Human-in-the-Loop and highlights how LLMs can support monitoring, analysis and planning, execution, runtime models, and human collaboration. Key contributions include a taxonomy of LLM-enabled SAS capabilities, concrete examples from SE, agents, robotics, and driving, and a roadmap of future research that includes integrating formal methods, collective intelligence, and runtime reasoning. The study provides a benchmarked snapshot of potential LLM applications in SAS and emphasizes the need for systematic, deeper investigations to translate potential into reliable, real-world solutions.

Abstract

Large Language Models (LLMs), with their abilities in knowledge acquisition and reasoning, can potentially enhance the various aspects of Self-adaptive Systems (SAS). Yet, the potential of LLMs in SAS remains largely unexplored and ambiguous, due to the lack of literature from flagship conferences or journals in the field, such as SEAMS and TAAS. The interdisciplinary nature of SAS suggests that drawing and integrating ideas from related fields, such as software engineering and autonomous agents, could unveil innovative research directions for LLMs within SAS. To this end, this paper reports the results of a literature review of studies in relevant fields, summarizes and classifies the studies relevant to SAS, and outlines their potential to specific aspects of SAS.

Figures (1)

• Figure 1: Overview of Categorization. The numbers in brackets indicate the number of studies in the category. One study can be categorized into multiple categories.