The AI-driven Manufacturing Transformation Is Imminent


Artificial intelligence, smart manufacturing and digital technologies are radically changing the competitive landscape across industries. As end users generate ever higher demands for personalised products, manufacturing is becoming more complex and fragmented than ever before. 

The challenge that organisations face is how to adapt manufacturing systems which are traditionally designed for high volume to accommodate mass customisation. Can the application of AI in the manufacturing context be the solution? Inexhaustible computing power, sophisticated algorithms and big data sets have already propelled the use of AI technologies in many sectors including healthcare, education, autonomous driving and retail. 

In addition to the industries aforementioned, manufacturing also has an abundance of data that traditional IT systems cannot utilise. As a production manager in the electronics industry, I have made several observations on opportunities for AI in the factory setting.

Yield improvement has long been the number one task that I have had to address. Rather than just relying on "experience," can AI make sense of data from raw materials, equipment settings, ambient parameters, past yield performance and advise on how to set up machines to generate optimal yield? Can AI reduce quality defects? What could AI do on production planning, as Excel and traditional planning systems are inflexible and cannot learn from past variables? Can AI diagnose machine failures and pin-point specific issues instantly so that when a technician is alerted they know which spare parts to bring and can minimise repair time?

I have seen a growing number of AI cases in manufacturing in recent years across the world. Right now, a major LCD panel manufacturer in China is using AI to detect and classify over 120 types of defects. Initiating as a pilot project in their Array factory in 2017, they are now able to reduce approximately 70% of human inspection labour and cut down more than 85% of inspection lead time. They are now implementing the AI programme across Cell and Module factories and are in the process of building a corporate AI platform. The intention is to extend well beyond inspection as they have experienced a glimpse of the power and true potential of AI in manufacturing.

In Germany, a machinery manufacturer has empowered their field technicians with a mobile AI system. Having analysed all types of structured and unstructured data from previous repair records, user menu, external technician forums and other sources of machine operating/sensor data, the AI system can assist technicians to reach average repaid time reduction by 45%.  

AI is also helping a global electronics manufacturer’s supply chain. It connects with internal and external IT systems and advises supply chain managers on possible inventory shortages, late shipments and how to reduce the impact of such disruptions by using resolution rooms to manage and control key supply chain activities.

Many of these AI applications cases have originated from just one AI system addressing one particular problem. As business/data flows are designed to be connected in manufacturing, these AI systems are also becoming connected with each other. The AI system which reads product quality in real time also sends the output to another AI system that performs yield analysis and uses the same data feed to diagnose and monitor equipment performance. A significant drop of yield will be received and interpreted by supply chain AI systems which then advices supply chain managers to use alternative factories that have extra capacity to fulfil customer orders.

This connected “Hive” like AI network is in the process of transforming how manufacturing operates. Overtime, it will become a centralised and powerful “corporate brain”. Production managers, technicians, purchasers, R&D engineers, supply chain managers and C-suite executives will all be assisted by such corporate AI system in decision making and daily activity management. Cross-organisation collaboration can be more effective through connection of different corporate AI systems as exchange of intelligence is far more valuable than exchange of data.

We are at the dawn of this new era for manufacturing. Together, with new technologies such as blockchain, quantum computing, smart contracts and other digital capabilities, AI will not only change how we operate manufacturing, but also transform and disrupt business models entirely. Be prepared.  


Qin Deng is Global Electronics Industry 4.0 Solutions Lead at IBM

Daniel Camara