By 2035, it is projected that artificial intelligence (AI)-powered technologies could increase labor productivity by up to 40 percent across 16 key industries, including manufacturing, according to a recent report by Accenture and Frontier Economics(1). In dollars and cents, that could equate to a $3.8 trillion increase in economic output in the manufacturing sector alone, or an almost 45 percent increase compared to baseline estimates. With that degree of growth potential, it is clear that AI has become – and will remain – an integral factor in manufacturing.
Because of its dependence on heavy machinery and equipment, manufacturing is considered a capital-intensive industry. And as such, it is particularly well-suited to the application of AI technologies.
“AI will perform manufacturing, quality control, shorten design time, and reduce materials waste, improve production reuse, perform predictive maintenance, and more,” says Andrew Ng, Google Brain creator and computer science professor at Stanford University(2).
The concept of AI, or machines being able to carry out tasks and processes based on previous inputs or data, has existed since the 1950s. It replicates the way the human brain works with respect to memory, perception/sensing, learning, identifying patterns, and performing mathematical calculations. It follows, then, that as our understanding of the human brain evolves, so does our conceptualization of AI. And as that happens, new applications in the manufacturing space emerge. This article provides a snapshot of just a few of the latest ways that AI is changing manufacturing.
A time-consuming yet critical task found in most manufacturing environments is the visual inspection of produced goods to ensure they meet quality specifications. Usually, this involves people manually inspecting samples taken from a production batch or lot. Depending on the size and nature of the goods produced, workers are often assisted by microscopes and other technology to more easily spot defects. Deviating product is then passed to quality managers to assess severity and scope, triage, investigate root cause, and determine proper disposition.
As with any human endeavor, at least some degree of error is inevitable in this context, and to an even greater degree, inefficiency. But a Silicon Valley startup formed by Ng has developed computer vision tools that, with the help of a machine learning algorithm trained on relatively few sample images, are able to detect even the most minute imperfections. These tools are not only able to spot issues quickly, but they can also report them for further processing, an AI process called automated issue identification.
Computer vision bodes well for overall quality in manufacturing. With AI-assisted vision capabilities, quality issues can be identified and addressed faster and at the source, allowing for more compliant product that’s right-first-time and ready to ship the moment it leaves the production line. And because this technology is also able to capture and report issues in real time in an electronic system, a company can feed this information into the audit trail and perform advanced analytics to prevent the issues from happening again.
According to recent studies, unplanned downtime costs manufacturers upward of $50 billion a year(3). It’s no surprise, then, that one of the most promising use cases of AI in manufacturing is its ability to virtually eliminate both planned and unplanned downtime caused by faulty or idle machinery.
For years, manufacturers have used sensors and preventive maintenance to monitor and assess the performance of their equipment. But traditional preventive maintenance requires a fair amount of manual analysis of large volumes of data to produce any actionable conclusions – it therefore tended to be reactive. In contrast, because it is paired with AI technology, predictive maintenance can leverage advanced models and analytics to predict failures well before they occur. And over time, machine learning increases the accuracy of those predictive algorithms. This allows companies to realize the full potential of existing machinery on the factory floor while avoiding downtime altogether, resulting in optimized performance and avoiding the common, yet problematic, run-to-failure scenario.
When this data is made available to a company’s network of enterprise manufacturing solutions, key stakeholders such as maintenance and reliability professionals can access it – and take action – faster. This is just one example reinforcing the importance of fully digitized and integrated manufacturing systems, from enterprise resource planning (ERP), to equipment maintenance and calibration, and even production records.
The cost of deploying advanced technology in manufacturing continues to fall regardless of geographic location. Coupled with stabilizing cost of labor, manufacturers have greater freedom of choice in determining where to establish their production facilities to reduce the distance between raw materials, factories and consumers. Furthermore, pairing AI with other new technologies and capabilities is making manufacturing more responsive and agile, and enabling entirely new business models.
The ability of algorithms to spot patterns can be used in limitless contexts. For example, pattern-perceiving algorithms can identify consumer demand patterns over time, geography and market segments, while accounting for highly variable and unpredictable factors like macroeconomic cycles, political developments and even weather(4). Already, algorithms have been shown capable of interpreting the sentiment in social media feeds precisely enough to predict changes in the stock market. Researchers foresee these same algorithms can be applied to the dialogues and interactions consumers have with their smart appliances and assistants to project demand for specific products or brands.
Combined with technologies like 3D printing, centralized quality control, digitized product design, the industrial IoT and cloud computing, AI is enabling the concept of “batch size one” or “mass customization,” where companies co-create with consumers to produce demand-specific, highly personalized goods efficiently and at high quality. Doing so will not only become economically feasible, but even profitable for companies. Both Siemens with its Click2Make solution and GE with its Brilliant Manufacturing Suite are innovating in this space(5).
While words like transformative and disruptive are commonly used to describe the current technological trends in manufacturing that are collectively referred to as Industry 4.0, the fact is that change has always defined this industry. Manufacturing was born of the major technological advances that came about during the First Industrial Revolution in the late 1700s, and has been shaped and influenced by further advances ever since.
But interestingly, the promise of AI is not exclusive to the context of machines. By automating repetitive manual tasks and streamlining time-consuming data analysis, AI will serve to augment the human workforce and allow people to be redeployed to more critical, innovative tasks. Industry 5.0, the next chapter in the evolution of manufacturing, is based on the very idea of emphasizing the human element in manufacturing and enabling humans and technology to collaborate as opposed to working disparately.
According to Paul Daugherty, chief technology and innovation officer at Accenture and co-author of the AI report referenced earlier in this article, “To realize the opportunity of AI, it’s critical that businesses act now to develop strategies around AI that put people at the center, and commit to develop responsible AI systems that are aligned to moral and ethical values that will drive positive outcomes and empower people to do what they do best—imagine, create and innovate.”
So when it comes to manufacturing and AI, it seems the story is far from over. Stay tuned.
Enjoying this blog? Learn more.
The Ultimate Guide to Connected Quality DataGet the Guide