Humans and AI in Manufacturing: mission impossible?
The World Manufacturing Forum 2020 Report says: “Artificial Intelligence is not novel in manufacturing. In the last decade however, thanks in part to advancements in AI algorithms, computational power, connectivity, and data science, it has gained more importance as companies increasingly see it as a driver for competitive advantage. However, the lack of experienced talent to work with AI, lack of know-how, and the need for accurate data remain important challenges for organisations in adopting AI.” Unlike other more recent and greenfield application domains mostly related to citizens and B2C, the business benefits for AI adoption in Manufacturing needs to be contextualised in the 6Ps dimensions of an Enterprise. How AI-driven Digital Transformation affects company’s Products (e.g. connected cars), Processes (e.g. maintenance), Platforms (e.g. MES ERP PLM systems), People (e.g. professions and skills), Partnerships (e.g. Digital Innovation Hubs and SMEs) and Performance (e.g. twin transition digital-green business indicators).
More specifically, speaking of the People dimension, WMF2020 report says “Currently and even more so in the future, people are consistently interacting with AI. As a result, this changes the nature of work and technological interaction. Particularly, the modern workplace will be affected by the implementation of more AI into processes.” All research and innovation initiatives aiming at introducing AI in manufacturing need to address the challenge of human adoption of increasingly AI-based autonomous systems. New roles and professions, new skills and competencies need to be developed and adopted in the Factory, in the Product Design and Engineering Departments, in the Value Chains.
But how does the Human-Machine interaction happen? How to model it, how to simulate it and how to improve and enhance it in the workplaces? In 2018, Harvard Business Review published an article entitled “Collaborative Intelligence: Humans and AI are joining forces”, envisaging a Human-to-Machine train-explain-sustain and a Machine-to-Human amplify-interact-embody interactive processes. As a matter of fact H-to-M interaction is like a Parent-Child interaction where more experienced and knowledgeable beings are able to train about processes and procedures, to explain why certain decisions have to be taken and to sustain autonomous systems in the accomplishment of ethical, legal, behavioural and governance issues. The M-to-H interaction is similar to the relationship Caregiver-Elderly where more capable beings need to amplify physical or cognitive capabilities of less able people, need to enhance their interaction with other beings or data, need to embody their knowledge and capabilities like in a robotic arm or in an exoskeleton.
According to the EU Commission, Industry 5.0 “attempts to capture the value of new technologies, providing prosperity beyond jobs and growth, while respecting planetary boundaries, and placing the wellbeing of the industry worker at the centre of the production process.” In the Industry 5.0 perspective, next generation AI systems need to consider human factors and human acceptance as a key competitive advantage for their adoption in Manufacturing Industry. Collaborative Intelligence workplace models, simulation and operational systems is one significant way to achieve Industry 5.0 objectives and a more human-centric and sustainable factory of the future.
The STAR ICT-38-2020 project is working in this direction, focusing on leading edge AI technologies with wide applicability in manufacturing environments, including Explainable AI, Active Learning, Simulated Reality, Human-Centric Digital Twins for AI, security and trust for AI systems in manufacturing. The STAR marketplace will be soon available for manufacturing SMEs to grasp and adopt the most recent "AI for Manufacturing" solutions also in the field of Human-AI Collaborative Intelligence. STAY TUNED!!!
By: Andrea Ferretto Parodi, GFT Italy