The Rise of AI in Manufacturing

As tariff policies change rapidly, companies are scrambling to understand the volatile impacts on their operations, from raw material costs to supply chain disruptions.

To better navigate this, software developers have prioritized making digital tools to help manufacturers and retailers have better visibility into how tariff fluctuations are affecting them.

Avanade, a joint venture between Microsoft and Accenture, created its Tariff Navigator tool to provide real-time alerts, cost updates and insights that business leaders can use to make faster, more sound decisions.

Tom Nall, global manufacturing and automotive industry executive at Avanade, presented the Tariff Navigator during the North American Manufacturing Excellence Summit in Fort Worth, Texas, in April 2025.

Instead of searching through records and systems to understand what tariff codes are in place, he said users can interact with Avanade’s tool, which uses artificial intelligence to comb through the company’s data, by asking it simple questions and requests.

In a video demonstration, a procurement manager named “Alex” is able to get real-time alerts of tariff changes, update SKU classifications in seconds and see how purchase orders are affected by suppliers.

The tool can also generate reports highlighting the top items affected by tariffs to help him begin to explore alternative suppliers or distribution options, according to the company’s presentation.

“He never once had to go into an ERP system,” Nall said. “He never once had to go out to a governmental system that keeps the records and the tariff changes. He never once had to go to his IT department and say, ‘write me this report and send it to me’ or download things to Excel to manipulate it.”

“All he did was talk to a system,” Nall added. “All he did was ask that system everyday questions, and it was able to go behind the scenes and give him the information that he needs to do his job.”

Enterra Solutions, a data analytics firm that works with food manufacturers such as Nestlé and General Mills, launched its own AI tariff tool for data-backed decision-making assistance.

The Enterra Dynamic Enterprise Resiliency System is able to anticipate market shifts and recommend actions to take with 91% accuracy amid volatile economic and political environments, according to a news release.

“Legacy systems, technologies and processes were built for a time in which seismic economic and geopolitical changes happened every decade, not every month or day,” Stephen DeAngelis, founder and CEO of Enterra, said in a statement.

“We can now power business leaders to quickly understand what levers they can pull and what decisions they can make to turn risk into opportunity,” DeAngelis added.

Beyond manufacturing, Edited, an AI-powered retail intelligence platform, has launched its new tariff tracker to help retailers respond faster to tariff impacts. A version of it is available to the public and can show how tariffs are shaping pricing, assortment and promotional strategies in real time.

“With Edited, they get real-time insights and AI-driven recommendations to move fast, protect margins, and stay ahead of the market,” Shellie Vornhagen, chief marketing officer at Edited, said in a statement.

Edited’s tracker draws from more than 5 billion SKUs across 5,000 retailers and more than 14 years of history, according to a news release. Retailers can also make agile pricing strategies, smarter inventory planning and spot inventory risks early through Edited’s full retail intelligence platform offerings.

Chemical manufacturers are increasingly turning to artificial intelligence to enhance operations.  

AI usage in the chemical industry has particularly increased over the last five years, Seneca Fitch, director of the health sciences practice at scientific consulting firm ToxStrategies, said at a GlobalChem conference panel in Washington, D.C. in April 2025.

Fitch was joined by Rebecca Morones, a senior product steward at BASF, and Sean Watford, an environmental systems and information scientist at the U.S. Environmental Protection Agency, to discuss how they use AI and how to train the technology to give the answers needed.

AI can be used to perform risk assessments, as well as simplify and summarize laws with hundreds of pages, such as the Toxic Substances Control Act, to help companies stay compliant.

Data management, Fitch added, is an important use of the technology, but only if manufacturers start with quality data. 

“Garbage in equals garbage out. We've heard that saying a lot, and so it's very important that we are using optimal data when we are training models, so that we are making sure to get the most valid and reliable results,” Fitch said.

AI hallucinations can skew results

Morones said she uses Microsoft Copilot to summarize information, such as EPA risk evaluations. She noted, however, that while the platform provides highlighted outlined points, it does not perform an in-depth analysis of information, Morones said.

“You still have to read the documents,” Morones said. “But I feel like there's also 200 pages it's sifting through, which does at least help get you to kind of where you're looking for and give you a great oversight.”

However, AI can pull incorrect data, also known as hallucinations, Morones added. Her colleagues have asked AI for the density of a safety data sheet or what a TSCA fee is, and the platform found an incorrect answer.

“So if your businesses are using this, and I do caution that, just make sure that they are aware of these hallucinations,” Morones said. “And anytime it comes to regulatory aspects, you should probably have them come to the regulatory experts. Don't rely on what they're seeing or what they're doing.”

At other times, the technology may pull up references that appear legitimate but actually do not exist, Fitch said, adding that any information generated by an AI model should be independently verified by staff to ensure its validity. 

“With that lesson, it's really important that not only are we reviewing the things that are coming out of any AI model, but also that we're verifying them,” Fitch said. “Because something could certainly sound logical, it could certainly look real, and yet it's not. It's false information and that hallucination, that's not rare.”

Leveraging prompt engineering

To obtain the necessary answers to queries, manufacturers will need to use prompt engineering to train the technology.

“I think that the most practical advice is that it is a computer and it uses logic,” Fitch said.

Morones described prompt engineering as instructing a computer to perform a simple task, such as how to make a peanut butter and jelly sandwich. “It's not going to understand, ‘Grab two slices of bread,’” Morones said. “You have to be very specific for what you're doing.” 

One strategy the EPA uses to reduce hallucinations through prompt engineering is to ask the model the same question multiple times, to ensure consistency in answers, Watson said. “If the question or the task might be repeated over and over again, you want to ensure that you're getting similar performance for each iteration of the task,” Watson said. 

If companies are interested in exploring AI for regulatory purposes or efficiency, Morones said qualified employees should be involved in its development.

“You’re the one that understands better than AI is going to understand,” Morones said. “You know what you're looking at, what you need to find.”

Sponsored content

By Wiley

Companies who want to leverage AI in high-stakes decision-making can improve outcomes by integrating trusted, high-quality content alongside proprietary data in AI applications.

AI is already reshaping how the world discovers knowledge and generates insights to drive faster decision-making and improved outcomes.

In today’s rapidly evolving AI landscape, the differentiating factor between success and failure increasingly comes down to a critical element: the quality of content used to train, inform, and refine AI models.

Researchers, developers, and organizations that integrate structured, high-quality content into their AI workflows are gaining a clear competitive advantage. 

Information drives innovation

Innovation has always been powered by the ability to access the most current information. While physical libraries once served this need, today’s information landscape offers digital access alongside an exponential volume of content. This abundance has created challenges and sparked new approaches to organizing, cataloging, and engaging with information by individuals and teams across industries. 

Generative AI (GenAI) allows organizations to transform their production workflows, enhance data analysis, summarize information, translate content, and automate tasks. But successful implementation requires more than just letting AI handle the heavy lifting.

Investment in data infrastructure

Organizations that make strategic investments in building, curating, and effectively leveraging such distinctive data assets will be best positioned to achieve superior AI performance, develop innovative AI-driven products and services, and realize tangible business outcomes.

To ensure accuracy and relevance, GenAI applications must be built on a foundation of high-quality, well-structured data, including both internal proprietary data and trusted external resources.

The role of third-party data

Research indicates that as many as 85% of AI project failures can be attributed to issues related to poor data quality. 

While proprietary datasets offer a wealth of quantitative and company-specific insights, third-party datasets add a layer of verified data that capture intelligence from the wider industry to expand and strengthen the insights an application can pull from in its synthesis.

Published scientific content is a category of third-party data that has enormous potential to accelerate innovation and capture supply chain insights across a range of research-intensive and highly regulated industries. Scientific content may include journal articles, systematic reviews, textbooks and reference works, conference proceedings, and other technical or professional practice documentation.

Journal articles and conference proceedings deliver cutting-edge insights from the latest advancements in a field of study. Textbooks and reference works give applications access to established consensus in a field of study and an authoritative source for the current state of knowledge in a specific domain.

Data you can trust

AI models, by their very nature, are reflections of the data they are trained on.

Authoritative and accurate training sources are vital. But how can you be sure that scientific content is authoritative and accurate?

Journal articles in particular go through rigorous editorial processes, including transparent and documented peer review workflows. Rigorous standards are applied to ensure that only factual and methodologically sound research makes it into the scientific record.

Other types of content, like textbooks and reference works, are developed with close editorial oversight from leaders in the field and are written by authors who are well-respected domain experts in their own right.

Working with publishers like us means your AI teams can also benefit from publishers’ use of structured metadata with comprehensive taxonomies and classifications that are built into the data.

How to get started

Companies with robust existing AI infrastructure who are ready to move beyond proprietary datasets should start by understanding the types of content that can drive a particular use case forward. Work with end users and subject matter experts to understand what information sources they rely on for reference and support in their day-to-day work.

Once use cases have been identified, organizations seeking robust scientific research to drive applications forward can explore licensing options for the Wiley portfolio, which spans all major fields of study with well-respected books, journals, and other types of robust scientific content.

As one of the world’s largest scientific publishers, we see AI as an extraordinary opportunity to accelerate innovation in research, learning, and publishing. Our approach to AI is grounded in four core principles to promote ethical, innovative progress to drive discovery.

Models are everywhere; data is the differentiator that will help companies beat the competition and accelerate innovation and success.

Manufacturers are implementing artificial intelligence into their processes to boost efficiency and gain a competitive edge.

Eventually, with the right combination of robots, data and software, it could lead to fully autonomous semiconductor fabs or processes that free up the human workforce to solve problems alongside AI, according to executives at an industry conference in May 2025.

Leaders from Intel, EMD Electronics, Global Foundries and other computer chip companies discussed their visions for AI at the Advanced Semiconductor Manufacturing Conference in Albany, New York in May 2025. They also spoke about the challenges and limitations facing the technology as adoption grows across the industry.

From issues around data scarcity to hallucinations, here are some of the key limitations around AI that industry leaders are facing.

Identifying where to get value from AI

Currently, tools like ChatGPT and other large language models can generate human-like text and perform language-related tasks. Some tools can generate video, images or even code, while other forms of AI can create digital replicas of factory floors, handle repetitive tasks or improve quality control through computer vision.

There are countless possibilities for the use of AI in chip manufacturing. However, as companies race to incorporate the technologies into their operations, they can fall into the trap of implementing tools without understanding how they can improve performance.

“People tend to use that hype. Everything is AI. Let's try to do something with AI, but we don't get value out of how much we invested,” Safa Kutup Kurt, global head of plant operations and digital transformation at EMD Electronics, said during a conference panel. “We need to find that balance, and we need to really work on scalability of the solution and generating the value.”

“If we cannot quantify it, well, it's going to be difficult to scale,” Kurt added.

The ‘explainability’ of AI

AI tools can provide answers to solutions, but understanding why they came to their conclusions is tricky.

“Improving just the explainability and building trust in these models is a huge way…that they need to evolve,” Jason Komorowski, senior automation and analytics engineer at Intel Corp., said during the panel.

“We can't just give people a black box and say you input what you want and it'll spit out what you need to do for it. Right? We have to be able to explain it, understand which features are being used, and how we're building and coming to our decisions,” Komorowski added.

Improving partner collaborations on AI

Companies tend to focus on advanced technology improvements within their own organizations. Those sorts of issues are easier to fix, said Pawitter Mangat, vice president of global tapeout and mask ops at Global Foundries. It’s getting the companies aligned with their partners on technologies across the entire end-to-end market that is most challenging, Mangat added.

“We all have different processes, but we have different models,” Mangat said. “How do you optimize that from that perspective, to say we don't have a customized solution for [Global Foundries] versus Intel versus other companies, right? That's where I think the biggest challenge is, and we can all gain if we have a process that allows us that shading, or at least define the baseline that we're aligned in some of these models.”

Data scarcity

In addition to increased collaboration, data availability and accessibility is going to be a challenge for progress and innovation.

AI models require vast amounts of data to operate effectively without variance, and often companies are not willing to share proprietary data with others, Ross Kunz, data scientist at Idaho National Laboratory, said during the panel.

“Not one single model will actually go ahead and provide you the best results that were right,” Kunz said. “ChatGPT is not going to provide you the best results for every single process. And so there has to be some sort of robustness associated with models such that we can actually compare across different industries and develop … methodology such that it can be applied across multiple scales.”

Validation of AI models

A lot of progress has been made to get AI models functional, but data validation remains a challenge.

“We read in the news all the time about hallucinations with Gen AI systems, right?” Komorowski said.  “It's very hard to validate if the answer is correct or not without seeing up with a person saying, Yeah, this makes sense. So validation of these models is certainly a limitation.”

While progress around AI has been made to improve shop floor operations, such as the creation of digital part inventories to reduce waste and running full factory simulations, there is more work to be done to get the most out of the advanced technology.

More research and development is needed to ensure the models can generate more deterministic results rather than probabilistic results, Komorowski said. He also underscored the need for improved AI “explainability” and cross-collaboration between vendors and customers.

As progress is made to create standardized platforms and formats for sharing, Komorwoski said that opens up the way Intel and other companies can use AI across the factory floor “so we’re not having to recreate it every time, and that’s a huge enabler.”

Siemens AG acquired Troy, Michigan-based Altair Engineering in an effort to expand its industrial software portfolio and artificial intelligence capabilities, the automation manufacturer announced in March 2025.

The German multinational group said it completed the $10 billion deal, allowing Siemens to leverage Altair’s advanced simulation technology for its digital twin tool Xcelerator. 

The acquisition is part of Siemens’ “One Tech Company” program launched last year with the aim of growing its digital revenues through investments in AI-enabled products and connected hardware.

As manufacturers look to embed advanced technologies into their day-to-day operations, Siemens is positioning itself as a major resource with extensive AI-powered tools.

The acquisition of Altair will strengthen Siemens’ industrial software business, which has been on a growth trajectory since 2007, President and CEO Roland Busch said at a November 2024 press conference. It also underscores the company’s commitment to its “One Tech Company” program focused on improving internal processes and accelerating digital investments.

“With its full-suite simulation and AI portfolio, Altair will boost our comprehensive digital twin to the next level,” Busch said in remarks. This is Siemens’ largest acquisition to date, he said.

Altair will expand Siemens’ digital offerings for its customers with new capabilities in mechanical and electromagnetic simulation, high-performance computing, data science and AI, according to a news release.

“All our businesses are working to expand and promote the Siemens Xcelerator Portfolio,” Busch said in November 2024. “The integration of Altair Engineering will further boost the share of our digital revenue.”

In addition to Altair, Siemens is targeting smaller businesses with the acquisitions of Trayer Engineering Corporation, a California-based switchgear manufacturer, and Denmark-based fire suppression firm Danfoss Fire Safety.

The multinational company is also selling its airport logistics business to Toyota Industries subsidiary Vanderlande and is looking to divest most of its wind business in India and Sri Lanka. It opened two facilities in March 2025 in California and Texas as the company focuses on AI-related investments in the U.S. 

Altair, founded in 1985, began as a startup consulting firm for Detroit’s largest automakers, Crain’s Detroit business reported. It has since grown into a leading software and technology company with an emphasis in simulation, data science and high-performance computing. The company went public in 2017.

As President Donald Trump’s administration modifies tariff regulations daily, consumers are shifting their purchasing power in an effort to cut costs amid an uncertain global climate.

A lot of companies are struggling to balance innovation with the economic realities of inflation and supply chain uncertainty, which increased with pending tariffs, Ivan Torossian, consulting director at GlobalData, said during a virtual State of Innovation in Food Manufacturing event hosted by Manufacturing Dive and sister publication Food Dive on April 9, 2025. 

Torossian was joined by Megan Bullock, predictive science director at Conagra Brands, and Franz Guzmán, a consultant and marketing professor at California State University, to discuss how food manufacturers can navigate through these challenges to continue developing new products and processes.

As the global trade war rages on, Torossian said food manufacturers will need to come up with a strategy to not only make the consumer happy but also make it worth their money. 

“We don't know what's going to happen tomorrow, but basically they want and need to innovate, but are facing this intense cost pressure,” Torossian said. “So many are concerned about how to maintain their innovation pipelines while managing these margins in this environment.” 

While innovation has changed over time, there are three components that are always present, Bullock said.

“You have the factor of convenience and value, you have the factor of health, and you have the factor of flavor and experience,” Bullock said. “And those will always be standing pillars when it comes to building a new product. But the nuances underneath those have shifted a lot.”

The uptick of social media and viral trends have also led to consumers wanting bold flavors that are buzzworthy, she added. 

“Social media is one of our core data sources because it’s a real-time feed of what consumers are cooking and eating and dabbling with, in home and at restaurants, but there is a lot of noise,” Bullock said. “You have to decipher between what people like to watch versus what they actually want to recreate.”

The traditional stage-gate model for product innovation cannot keep up with the rapidly changing consumer interest on social media. As a result, companies are rethinking their innovation processes to close the gap, Torossian said. Moreover, companies are conducting “renovations” of existing products or processes to generate quick returns.

“I think that leading companies like Conagra are adopting what we call adaptive innovation, so creating more flexible platforms that can quickly pivot as these trends evolve,” Torossian said.

Companies like Conagra are utilizing artificial intelligence to accelerate innovation, which takes a fraction of the time, from six months down to six weeks, and is cost-effective. Instead of getting consumers in a room to taste test the products, Conagra looks at behavioral data, Bullock said. The data includes what consumers are searching for on the internet, what they're talking about on social media, and what kind of products they're buying.

For example, the consumer demand for GLP-1 weight loss drugs, which has been changing food consumption, according to a February 2025 EY report. Conagra didn’t immediately jump on the GLP-1 trend. Instead, the company watched the demand increase progressively before modifying the packaging to say it was GLP-1 friendly.

“If there's new development in terms of certain ingredients or products that work for them don't work for them, and we might get to a tipping point where we will introduce new solutions,” Bullock said. “But we're just naturally always going to follow that approach, where we inch in first by making that better connection before we jump right to innovation.”

However, it can be too much to sift through disparate datasets that Conagra has to connect and find similar patterns, Bullock said. With AI, however, the technology can go through the data that helps the company come up with ideas, such as what kind of color cue to use for a certain attribute.

Torossian said he sees AI revolutionizing food innovation across the entire development process, analyzing data points to identify emerging consumer trends before they go mainstream. 

Formulation-wise, AI optimizes variables, such as taste, texture and costs, he added.

“AI will enable brands to create products tailored to specific consumer sectors or even individuals,” Torossian said.

While the amount of data can be overwhelming, and there are challenges to using AI, such as regulatory and legal hurdles, Guzmán said to play with the technology to ensure companies understand it.

“I need you to know what I can do with that data for more strategic thinking if we position ourselves in that way,” Guzmán said. “It’s not gonna take our jobs, but it actually requires us to level up our game.”

IBM has developed and implemented an artificial intelligence-powered PFAS screener tool that helps identify and eliminate fluorochemicals from its research operations, according to a June 13, 2025 blog post.

Dubbed the Safer Materials Advisor, the technology also suggests alternatives the company can use in place of “forever chemicals,” if there are viable substitutes available.

While the screener cannot guarantee the product is PFAS-free, it reduces errors and frees up IBM employees’ time to do other tasks, Angela Hutchinson, a chemical coordinator at IBM Research’s headquarters in Yorktown, New York, said in the blog post. 

Hutchinson is tasked with reviewing chemical requests related to research being conducted at IBM’s internal lab. Researchers within the company submit chemical requests linked to research IBM’s lab, from semiconductors to quantum computing.

Usually, chemical coordinators must review each chemical request, which consists of reviewing a substance safety data sheet. The sheet contains a chemical’s data, including toxicity, procedures for spills and leaks, storage guidelines, first-aid and firefighting measures and regulatory information, according to the University of California San Diego. 

They must also check that the chemicals are not restricted or banned within the company, or on the local, state or federal level.

The Safety Materials Advisor can put the chemical through up to three screenings to provide the same information, which also helps find alternatives for other toxic substances. The AI-powered tool can be adapted to new hazardous chemicals.

IBM Research has been leveraging AI and advanced computing for years to make data processing more sustainable, in part, through its AI family of models for materials research, according to the blog post.

The PFAS screener tool was accelerated with the launch of “PFACTS,” a collaboration led by IBM Research and involving Cornell University, Digital Science, Numat and the University of Pittsburgh. The ongoing project was funded by the National Science Foundation in December 2023 and developed a resource for anyone seeking to replace, redesign or remediate forever chemicals.

In March 2025, the group launched pfasID, an open-source screening tool hosted by ChemForward, a science non-profit that supports collaboration for safer chemistry. The web-based tool helps companies, researchers and policymakers simplify the process of identifying fluorochemicals, according to a March 2025 press release. The screener’s process includes assessing multiple PFAS definitions, generating reports for supply chain partners as well as access to the company’s entire supply chain.