AI in manufacturing will shift from edge-case utilization to vertical integration
#AI22 is a series of articles highlighting what we believe to be 10 developments that will be impacting AI this year.
This series is co-written by Dr. Johannes Otterbach, Dr. Rasmus Rothe and Henry Schröder.
The realization that AI will revolutionize the manufacturing industry is not new. However, the long continuing attempts of implementation have shown little widespread business use. While, by the end of 2020, 79% of manufacturers were piloting AI products, only 21% of manufacturers had active AI systems in production. Expect this to change as AI in manufacturing turns from edge case usage to mainstream business applications, super-charged by the effects the pandemic has had on the digitization of businesses worldwide.
It is evident that especially in an industry where the success of a business is largely influenced by the ability to repeat tasks quickly at high quality, AI technology provides great use. Monitoring, supply chain management and materials design are applications that will be revolutionized through AI. In the space of monitoring, the improvement of procedures such as predictive maintenance or quality control measures will experience widespread applications of AI. AI’s ability to process immense amounts of data, provided by the machines, from various data types and sources allows it to identify anomalies rapidly and can prevent the expensive downtime of machines. Additionally, McKinsey predicts that the forecasting ability of manufacturing through AI can be increased by up to 20%, thereby providing more efficient supply-chain management and inventory processes.
The third key application of AI will be experienced in the design of materials and products, especially in two situations: development of more sustainable products and development of highly complex products. Products that require numerous iterations or simulations to be designed, will be able to do so with the help of AI (e.g. digital twins). In computer chip design, for example, the application of reinforcement learning in simulating a large number of chip designs in combination with neural networks detecting the most efficient design for each use case will see early adoption by a few key players. Companies such as Nvidia believe these technologies will cut the time to produce computer chips in half while matching or exceeding a human-designed product.
Similarly, the application of AI generative design software will surge. Generative design is a technology where AI is applied to optimize designs based on the parameters input. In combination with 3D printing, this technique will become an exceedingly forceful tool in creating products more efficiently and reducing waste materials - without a decrease in performance. As companies are in the need to become more sustainable and are incentivized to produce more efficiently, generative design software will allow businesses to optimize any manufacturing process they desire.
While there are significant fields where the application of AI will substantially optimize the production processes, a major barrier continues to exist: scalability. Especially, SME’s have very customized factory and machine layouts, making it increasingly harder for AI models to scale rapidly into each factory, without significant customization of the systems. Two trends that try to circumvent this obstacle: firstly, the development of AI solutions that are mainly self-serviced - providing customers with the ability to adapt the systems themselves, without deep AI knowledge necessary within the company. And secondly, market players trying to become vertically integrated (e.g. building their own machine ecosystems). As the market addresses these challenges, expect significant application of AI in the manufacturing industry.