This publication explores the integration of Human-Centric Artificial Intelligence in the fashion industry, with a focus on the transition toward the Industry 5.0 paradigm. The objective is to optimize fabric cutting operations by reducing setup times and minimizing material waste. The main novelty of the work lies in the use of Large Language Models (LLMs) to translate operator feedback into quantifiable objectives for linear programming-based optimization algorithms. This work demonstrates how the fusion of AI and mathematical optimization can revolutionize production processes in the fashion sector.

The Fabric Cutting Problem

In the fashion industry, the cutting process is one of the most costly and complex stages. Traditionally, companies use commercial software to solve the cutting-stock problem, optimizing the placement of fabric pieces. However, this approach does not take into account the needs of human operators. Our research introduces a new mathematical model that integrates:

• Production data and logistical constraints
• Operator preferences and requirements, interpreted by a Large Language Model
• Multi-objective optimization to balance production efficiency and operator well-being

Human-Centric and AI Approach

The main innovation of this research lies in the integration of an LLM that interprets operator reports and translates them into quantitative parameters for the optimization model. The system allows operators to express feedback in natural language, which is processed to:

• Adjust cutting model parameters
• Minimize operator workload
• Improve collaboration between machines and workers

This solution represents a step forward toward a more flexible, adaptive, and human-centered industry, aligned with Industry 5.0 principles.

The analysis was conducted on a real industrial dataset, demonstrating that the proposed model:

• Reduces setup times compared to traditional methods
• Optimizes fabric consumption, reducing waste
• Improves productivity through the integration of human feedback