Successful AI adoption in manufacturing can’t be left to chance

By Randall Scott Newton

Manufacturing companies of all sizes are intrigued by the possibilities inherent in artificial intelligence (AI). A review of a recent survey on the subject leads me to suggest there are significant challenges ahead. 

According to data from Research and Markets, the market for AI in Manufacturing is expected to hit $20.8 billion by 2028, up from an estimated $3.2 billion in 2023. While past spending has been dominated by machine learning, future spending is likely to focus on predictive analytics, quality control, and context-aware computing. 

Those are lofty goals, but the reality of AI in manufacturing today is more sobering. Many manufacturing companies simply lack the necessary knowledge and skilled personnel to successfully implement and leverage AI solutions. IBM says 37% of manufacturing leaders surveyed cite this as a major barrier. Success strategies include building internal AI capabilities through training, hiring, and strategic partnerships.

The foundation of any effective AI system is access to large, high-quality datasets. Despite nearly a generation of PLM, ERP, SCM and other manufacturing-specific data technologies, many manufacturers still struggle with data being trapped in siloed systems across the organization. Integrating and cleansing data from disparate sources is a crucial step, say  31% of companies in the IBM survey. 

Beyond the data itself, 26% of manufacturers cite difficulties regarding access to the necessary software tools, as well as the hardware and digital infrastructure for developing, deploying, and maintaining AI models effectively. It looks as if AI lifecycle management will be the challenge of this decade. 

For smaller manufacturers especially, the implementation costs of AI could be prohibitively high. Between acquiring the AI technologies, integrating them into legacy systems, and supporting ongoing maintenance, these financial barriers remain an obstacle.

 The first generation of computers in the enterprise led to the acronym GIGA: “Garbage In, Garbage Out.” Manufacturing companies already generate massive volumes of data from IoT sensors and machinery. Robust data governance practices, secure infrastructure, and stringent data management protocols are required to feed AI algorithms with clean data insights and prevent GIGA all over again.

New technology always meets resistance from somewhere in the organization. Manufacturing leadership must have a focused change management strategy. Management at all levels need to address fears, provide ongoing training, and clearly communicate AI’s benefits.

Randall Scott Newton is managing director of Consilia Vektor. We resume regular publication with this posting.

Be the first to comment

Your comments are welcome