A generative pre-trained transformer industrial bot to improve operators’ working experience in a small Industry 5.0 factory

Abstract The industrial sector is embracing a new revolution, Industry 5.0 (I5.0), focusing on human well-being and ecosystem preservation. Previous industrial revolutions have emphasized incorporating the latest state-of-the-art innovations and technologies to improve production processes, overseeing some of their long-term damaging impact on human beings. Our research contributes to the I5.0 practical implementation in factories by proposing the design of a generative pre-trained transformer (GPT) industrial bot as a tool to assist workers of a small experimental factory in improving their daily working experience. The industrial bot closely tracks factory production and allows operators to monitor a few fundamental elements of their well-being via text queries, such as the pollution level and the maximum overtime hours per worker. It also provides factories with three valuable functions: faulty equipment discovery, root cause analysis, and synthetic data generation. These features originate from factory-customized data loaded into the OpenAI GPT4 language model (LM) presented in two industrial bot versions. The first is an offline version, created using a Langchain agent connected to the GPT4 LM through OpenAI APIs and visualized with Streamlit, an open-source front-end tool. The second is an online version produced by the GPT4 internal function that creates customs GPTs. We highlight both versions’ advantages and disadvantages to guide factories in considering the suitable environmental model when adopting I5.0. Our system also allows operators to receive additional tips, thanks to the online GPT4 robust database, to enhance their knowledge on maintaining their well-being and their factory contextualized on the query sent to the industrial bot.