Top 5 Thoughts on Implementing AI in Manufacturing

By Liz Stevens, writer, UV Solutions

Artificial intelligence is on the rise, again. The discipline began in the 1950s but fell out of favor in the 1970s. In the 1980s, AI became popular again, with expert systems that contained comprehensive knowledge for fields like law and medicine. Once more, however, AI fell out of favor. Now in its third incarnation, begun in the late 1990s and ramping up dramatically in the 2010s and 2020s, artificial intelligence seems to have gained sufficient traction and permeated enough industries to have staying power. Here, then, are a few thoughts on AI as it becomes a powerful tool in US manufacturing.

AI Implementation is Taking Hold

According to an industry survey conducted by manufacturing device and software maker Fluke Reliability, cited in Manufacturing Business Technology’s “Survey: Manufacturers Leading the Charge in AI Adoption,”1 implementation of AI is rising dramatically in US, UK and German operations. Of over 600 senior decision-makers and maintenance professionals surveyed in those countries, 90% of senior decision-makers and 80% of maintenance managers stated that AI will be a priority during their 2024-2025 calendar. When respondents who plan to invest in AI in the coming year were surveyed, some prominent benefits for implementing AI rose to the top of the list, including the technology’s usefulness in developing new products and services, its value in addressing data processing and analysis needs, its applications for improving customer service, its resources for upping efficiency and productivity, and the avenues it presents for tackling the skilled labor shortage.

  1. Generative AI Creates Content

Artificial intelligence has quickly gained acceptance among some users who create content like images, videos and text. Image creation tools like MidJourney and text generators like ChatGPT are standalone applications which tap into their own massive stores of art or language to serve up novel artwork or written output. In addition to creating output, artificial intelligence can be applied to analyze and interpret huge collections of seemingly unruly data. The Harvard Business Review article, “How Generative AI Can Augment Human Creativity,”2 describes how businesses can use AI – MidJourney, Stable Diffusion and ChatGPT, specifically – during the creative process to brainstorm and finetune innovative ideas.

“In the past two decades,” write the article’s authors, “companies have used crowdsourcing and idea competitions to involve outsiders in the innovation process. But many businesses have struggled to capitalize on these contributions. They’ve lacked an efficient way to evaluate the ideas, for instance, or to synthesize different ideas.” Generative AI can help ­overcome those challenges by supplementing the creativity of employees and customers to help them produce novel ideas — and improve the quality of raw ideas. “Specifically,” the authors write, “companies can use generative AI to promote divergent thinking, challenge expertise bias, assist in idea evaluation, support idea refinement, and facilitate collaboration among users.”

The article illustrates, with generated images and generated narrative, how these AI tools can take in a simple prompt and output a wide, unexpected variety of new possibilities. The tools also are valuable for their ability to create surprising output by combining images or reworking passages of text. And AI can be asked to evaluate ideas, for example, to assess the novelty, feasibility or impact of an idea. “Generative AI’s greatest potential is not replacing humans;” write the Harvard Business Review authors, “it is to assist humans in their individual and collective efforts to create hitherto unimaginable solutions.”

2. Artificial Intelligence Can Supercharge Data Analysis and Product/Service Innovation

Artificial intelligence is not a fix for every problem; it is a tool that should be applied where it is most suitable. In “Can AI Help Your Company Innovate? It Depends,”3 authors Lynn Wu and Sam Ransbotham discuss their research into how companies use AI for innovation and their finding that success in implementing AI depends upon how AI is used.

“We researched how companies are using AI for innovation and found that tools are just tools — success depends on how organizations use these new tools now at their disposal,” the authors write. “The results of these studies suggest that firms that have historically focused on specific types of innovation — process innovation and innovation by diverse recombination, in which companies combine a wide variety of technology elements in new ways — may benefit most from using advanced data capabilities of machine learning and AI.” The authors found AI to be less helpful as a tool for making incremental innovations or making small improvements to existing products. And they found that AI “is almost no help at all for radical innovation.”

The authors pose four questions to help companies decide if AI can be helpful to them in their innovation efforts:

  • “Are you a fast follower? Using AI can help improve upon existing products and create new products by recombining elements from prior successes. Advanced data capabilities will likely help, as you can use AI to amplify existing abilities in a new way.
  • Are you struggling with a deluge of data? While many researchers have deep expertise in one or two areas, emerging data technologies can help find technology elements from wide and distant fields. Broad synthesis is critical for these firms to combine these technologies to solve problems in their own domain.
  • Are you overwhelmed with choices? To succeed, you need to invest in AI projects that support recombination and hire AI talents to assess those suggested combinations. You don’t need to retrain or hire scientists with AI skills, but you do need to hire AI talents to support scientific efforts.
  • Do you depend on radical innovation? Using AI won’t likely directly benefit innovation efforts. However, using AI can lead to greater follow-up projects once you’ve made that radical first step.”

3. AI Can Improve Operations

While AI is a go-to tool for creating new content, making sense of massive data dumps or assessing the pros and cons of many combinations, it also can be a useful tool for identifying operations and process improvements. Writer Srinath goud Vanga points to some use cases for manufacturers to consider in the article “Harnessing AI to Streamline Manufacturing Processes.”4 Quality control and demand forecasting are just two of the areas that the author cites as strong use cases for AI.

To improve quality control on the factory floor, AI-powered computer vision stands out as a natural choice. “Human involvement in quality control can introduce inconsistencies,” writes goud Vanga. “AI-powered computer vision systems bring an objective eye to the process. These systems can meticulously detect even the most subtle flaws, ensuring uniform quality standards across factories and reducing the burden on human inspectors.” Companies might choose either to assist or to replace the quality control workers with AI. “These models can be deployed in two ways:” the author writes, “as a co-pilot to a human quality assessor, providing real-time feedback and highlighting potential issues; or as an independent system for highly repetitive and well-defined tasks.”

In demand forecasting, companies can turn to structured machine learning (ML) data and unstructured text for insights into future demand. Structured ML data is quite manageable but companies “often struggle to incorporate insights from the vast world of unstructured text,” writes the author. “This unstructured information that exists in the form of social media conversations, product reviews, and online trends have tremendous predictive power. The latest large language models (LLMs), such as GPT-4 and Gemini, have shown exceptional capabilities in converting textual data into quantifiable demand drivers and significantly improving forecast accuracy.”

And, according to the Industry Week article, “Easing into AI and Machine Learning,”5 it is a good idea to start an exploration of AI by applying it to edge rather than cloud processes. “Implementing AI and machine learning on edge devices closer to factory or fulfillment operations,” writes Elizabeth Samara-Rubio, “could provide solutions for the pressures manufacturers face with real-time computing and insights. By starting with targeted use cases to address decade-long innovation challenges, manufacturers can quickly demonstrate and gain value.” Samara-Rubio points out that AI and ML can help manufacturers optimize production scheduling, reduce waste and identify operation bottlenecks.

4. AI Attracts Workforce Talent

AI can conjure up novel images, make sense of feedback data overload or pinpoint production problems, but it can’t make job applicants line up at the door. Or maybe it can. In “AI Is the Key to Unlocking Manufacturing’s Future Workforce,” author Berk Birand promotes the value of AI in a manufacturer’s operations as a recruitment tool. “AI is transforming manufacturing from traditional, labor-intensive processes to high-tech, data-driven operations,” Birand writes. “This shift is crucial in attracting technically savvy college graduates who might otherwise overlook manufacturing as a career path. By integrating AI into manufacturing processes, companies can offer exciting opportunities in areas such as machine learning, data analytics, and robotics, aligning with the interests of tech-oriented graduates.”

Today’s young, bright workers also are interested in sustainability – an aspect of manufacturing that benefits greatly from tools like AI for waste reduction, energy efficiency improvements and thoughtful resource use. And new AI-based training technologies such as augmented reality and virtual reality are attractive to the younger generation and can be a safer approach for introducing new employees to production equipment. “The integration of AI also opens up new career paths within manufacturing,” writes Birand. “Roles such as AI specialists, data scientists and robotics engineers are becoming increasingly vital in the sector. These positions offer competitive salaries and opportunities for continuous learning and growth, making manufacturing a more attractive long-term career option for skilled graduates.”

5. AI Costs: Caveat Emptor

As with any technology, figuring the cost of a new tool like AI requires factoring in several areas of expense. There is hardware (including very specialized graphics processing units) and software, the cost of data models and their storage and management, and the cost to custom-train models (wherein enormous amounts of computing energy is required). There is the cost of ethical and regulatory compliance – areas that are still under development. For all of the above, there is the cost of skilled personnel – who do not come cheap. For an idea of costs shouldered by some of the biggest brand names in the business, a Time Magazine article, “The Billion-Dollar Price Tag of Building AI” states this: “Microsoft and OpenAI are reportedly planning to build a $100 billion supercomputer to build and run AI models.”7 And the cost for computing power is growing exponentially. Per the Time Magazine article, “The researchers found that the cost of the computational power required to train the models is doubling every nine months.” While a manufacturer contemplating an implementation of AI will likely purchase a cloud-based subscription or possibly an edge-based customized system with an industry-specific dataset, it is helpful to understand from where those seemingly high subscription or installation fees ultimately derive.

Artificial intelligence systems are relatively new computerized innovations; they have not entered the “economy of scale” period for a product or service. For some perspective, John Mark Suhy likens the AI landscape to the Wild West in the article “The true cost of AI: 6 factors government agencies should consider.”8 “The current landscape of AI mirrors the early, tumultuous days of the Wild West,” writes Suhy, “ – a period of exploration and untapped potential. As the U.S. government navigates this new frontier, the costs associated with AI projects are expected to be historically high, reminiscent of the early days of the internet or the space race. During these periods, initial investments were substantial as technologies were in their infancy, standards were non-existent, and the path forward was unclear.”

Artificial Intelligence: Lots of Input, Lots of Output, Lots of Smarts

Generative AI can create huge volumes of output, offering users a smorgasbord of creative ideas to weigh. Artificial intelligence can tease patterns or trends from massive collections of data. It can identify novel ways to combine existing ideas, designs and technologies. AI also can serve as an objective judge of new plans for products or services. It can imagine demand scenarios and predict future demand. It can be an aid to the quality control team, or it can replace the team entirely for standardized, repetitive quality inspection. Artificial intelligence can make manufacturing jobs more attractive to the technology-savvy generation of young adults.

AI is a tool that often operates in ultra-large volumes – interpreting huge amounts of diverse input or generating enormous quantities of novel or creative output. But AI also can be trained on quality, focused on a single idea or each emerging factory part, or tapped for its superior judging ability (informed by a voluminous knowledge base). This is the new, exciting, untamed and expensive world of AI.

References

  1. Industrial Media Staff. “Survey: Manufacturers Leading the Charge in AI Adoption,” Manufacturing Business Technology. June 12, 2024. www.mbtmag.com/artificial-intelligence/news/22911378/survey-manufacturers-leading-the-charge-in-ai-adoption.
  2. Tojin T. Eapen, Daniel J. Finkenstadt, Josh Folk and Lokesh Venkataswamy. “How Generative AI Can Augment Human Creativity,” Harvard Business Review. July-August 2023. www.hbr.org/2023/07/how-generative-ai-can-augment-human-creativity.
  3. Lynn Wu and Sam Ransbotham. “Can AI Help Your Company Innovate? It Depends,” Harvard Business Review. July 25, 2024. www.hbr.org/2024/07/can-ai-help-your-company-innovate-it-depends?ab=HP-latest-text-1.
  4. Srinath goud Vanga. “Harnessing AI to Streamline Manufacturing Processes,” Manufacturing Business Technology. June 4, 2024. www.mbtmag.com/artificial-intelligence/blog/22911656/harnessing-ai-to-streamline-manufacturing-processes.
  5. Elizabeth Samara-Rubio. “Easing into AI and Machine Learning,” Industry Week. March 21, 2024. www.industryweek.com/technology-and-iiot/emerging-technologies/article/21284980/easing-into-ai-and-machine-learning.
  6. Berk Birand, “AI Is the key to Unlocking Manufacturing’s Future Workforce,” Manufacturing Business Technology. July 11, 2024. www.mbtmag.com/artificial-intelligence/news/22914751/ai-is-the-key-to-unlocking-manufacturings-future-workforce.
  7. Will Henshall, “The Billion-Dollar Price Tag of Building AI.” Time Magazine. June 3, 2024. www.time.com/6984292/cost-artificial-intelligence-compute-epoch-report/.
  8. John Mark Suhy, “The true cost of AI: 6 factors government agencies should consider.” Federal News Network. June 25, 2024. https://federalnewsnetwork.com/commentary/2024/06/the-true-cost-of-ai-6-factors-government-agencies-should-consider.