Large Language Models in Industry 5.0 – Multi-Case Experiences from the AIMS5.0 Project

Hegedűs, Csaba and Olaszi, Péter and Bancsics, Máté and Ficzere, Dániel and Latipov, Alex and Holenderski, Mike and Strazds, Guntis V. and Abu Naim, Belal and Ghafourian, Yasin and Tauber, Markus and Bonik, Marta and Damkat, Chris and Dandolo, David and Masiero, Chiara and Barusco, Manuel and Susto, Gian Antonio and Almahasneh, Ruba and Varga, Pál (2026) Large Language Models in Industry 5.0 – Multi-Case Experiences from the AIMS5.0 Project. INFOCOMMUNICATIONS JOURNAL, 18 (SI). pp. 135-151. ISSN 2061-2079

Preview

Text AIMS_SpecIss_ICJ_2026_1_6.pdf - Published Version Download (2MB) | Preview

Abstract

The transition from Industry 4.0 toward the humancentered vision of Industry 5.0 creates new opportunities for AI systems that augment rather than replace human expertise. This paper presents the integration of Large Language Models (LLMs) and AI copilots into industrial engineering workflows through multiple cases within the EU AIMS5.0 project. Eight use cases span system-of-systems engineering, support and aftersales services, regulatory self-assessment, product engineering, factual correctness checking, robot programming, ontology and knowledge-graph applications, and AI-assisted testing. The cases range from conceptual designs and proofs of concept to pilot implementations and therefore provide implementation experience and design insights rather than directly comparable performance results. Across the cases, recurring architectural patterns include Retrieval-Augmented Generation (RAG), semantic search, multiagent planning, natural-language interfaces, code generation, and verification mechanisms. The experiences also reveal common deployment challenges related to heterogeneous and evolving data sources, multimodal content, access control, factual reliability, legacy-system integration, and evaluation. Viewed through the Industry 5.0 dimensions of human-centricity, sustainability, and resilience, the cases illustrate how LLM-based systems can support knowledge access, engineering assistance, system adaptation, and safer interaction with industrial AI. The findings emphasize that successful industrial adoption depends not only on model capabilities, but also on high-quality domain content, appropriate safeguards, transparent system behavior, integration with existing infrastructure, and continuous human oversight.

Actions (login required)

View Item