AI has long been an integral technology in manufacturing operations, but technological advances are opening up new opportunities to improve efficiency, safety and productivity.

The hype about an AI revolution in manufacturing is a myth, as the industry has been using automation and AI techniques for decades, according to the recently released Industrial AI Insights report from Honeywell Industrial Automation, a division of Honeywell that provides automation equipment, software and services primarily to process manufacturers. However, AI is evolving rapidly, and manufacturers are beginning to benefit from cutting-edge technologies such as generative AI.

For the report, Honeywell and Wakefield Research surveyed 1,600 leading manufacturing companies around the world that are currently implementing AI to find out how and why they are using these capabilities.

According to the report, AI projects in manufacturing are increasing even when the technology is not fully understood within the organization. Almost every respondent (94%) said that company leadership is fully committed to the technology, although 37% also expressed concern that their leaders do not understand AI.

Still, a large majority of respondents (84%) believe their organizations are AI pioneers, and 91% are finding new use cases for AI beyond initial plans to automate tasks and processes. Respondents also said their investments in their AI implementations are paying off, including increased efficiency through automation (64%), improved cybersecurity (60%), and real-time data to improve decision-making (59%).

There is also evidence that these new AI projects are still in their early stages: only 17 percent of respondents have fully launched initial AI projects, while others are still in the scaling phase (43 percent) or in the prototype phase (12 percent).

According to Jason Urso, CTO at Honeywell Industrial Automation, three types of AI tend to be used in manufacturing.

The most common form is deterministic AI, which has been around for decades and uses model-based prediction mechanisms to automate processes, Urso said.

Now, however, new forms of AI are entering the market that can bring new insights and benefits.

One is probabilistic AI, which uses large amounts of sensor data from operations and provides insights into the probabilities of what might happen. This can be used to monitor machines and determine when they may need maintenance, he said.

Finally, there is generative AI, which allows workers to “converse” with machines to determine their health status and suggest solutions to problems.

“This is a huge step forward because now (the machine) is communicating with the human in an almost human way, as if there was an expert sitting behind the scenes transmitting this information, when in reality it is your knowledge repository,” Urso said.

AI can help by reducing the time to acquire expertise. It doesn’t displace humans, but rather accelerates their acquisition of expertise by sending a GenAI query to a knowledge base that asks about a problem and can suggest solutions.

One of the biggest benefits of faster and better decision-making is that it will help companies better cope with growing labor shortages and skills gaps among younger workers in the manufacturing sector, he said.

“AI can help by reducing the time to expertise,” Urso said. “It doesn’t displace humans, but rather accelerates their expertise by sending a GenAI query to a knowledge base that asks about a problem and can suggest solutions.”

There are challenges that slow down full implementation of AI in manufacturing, he said, primarily around the quality of data and the ability of users to trust the answers they get from AI.

“We collect all the data, but can we trust it?” Urso said. “You have to make sure the dataset is accurate, otherwise it will give the wrong answer because of the faulty data. So it’s important to make sure there’s a thorough process to verify the dataset that’s used to make the recommendation.”

Despite the risks, AI is promising

The new possibilities of AI in manufacturing are promising, but companies must be aware of the potential risks of generative AI, according to a Forrester Research report published in March titled “Generative AI: What it Means for Smart Manufacturing.”

Benefits include improved business efficiency and effectiveness through generative AI applications that can help customer service representatives, knowledge workers and service technicians access a company’s growing data faster and get better answers, according to the report. The report’s conclusions are based on interviews with executives at several industrial companies, including Autodesk, Dassault Systemes, PTC, Schneider Electric and Siemens.

However, generative AI tools are not yet ready for widespread adoption in the manufacturing sector, says Paul Miller, an analyst at Forrester and one of the authors of the report.

“All the fuss about the risks of GenAI distracts attention from the good, useful and valuable work that manufacturers have been doing with other forms of AI for many years,” Miller said. “That would not be a good outcome for anyone.”

Manufacturers need to understand how generative AI differs from other AI technologies, he said. For example, an AI-based computer vision system that checks paint quality on an automobile assembly line is deterministic, or rule-based, because it produces the same output for a given input.

However, generative AI is non-deterministic, meaning the same input can produce different outputs, Miller said.

“This fundamental property of today’s GenAI tools will impact where and how you use them in the manufacturing context,” he said.

Many manufacturers are not explicitly aware of the non-deterministic nature of generative AI, but they are aware of some of the risks such systems can pose, such as the technology’s ability to induce hallucinations, Miller said.

“They are understandably concerned that a GenAI tool will give bad or incorrect advice that could lead to equipment damage, downtime or even injury, and they are keen to understand how techniques like retrieval-augmented generation can be used to limit the risks here,” he said, referring to a process that uses external data to improve the output of a model.

But manufacturers also want to know where the technology can add business value and are looking for guidance on how best to introduce the technology, among other options, to their employees and customers, Miller said.

Instead of trying to select an AI technology – deterministic or generative – and apply it to a business problem, Miller says manufacturers should focus on the business problem that needs to be solved.

“For example, saying ‘I want to spend less time searching for product documentation in field service use cases’ would be a far better starting point than ‘I want a GenAI tool for field service,'” he said.

AI in manufacturing is changing

The AI ​​revolution is nothing new in manufacturing, agreed Peter Zornio, CTO at Emerson Automation Solutions, an industrial automation systems and software company.

Manufacturing, particularly in the process industry, has been using digital systems to read and analyze sensor data and software to take corrective action since the 1970s, Zornio said. This has continued to evolve and is now leading to current machine learning and generative AI technologies.

“(Manufacturing was using) AI before AI was cool,” he said. “From the mid-1980s to today, we have been constantly applying these numerical mathematical AI techniques to achieve ever greater control and optimization of operations in the manufacturing process.”

The manufacturing industry has always had data, much of it coming from sensors that capture manufacturing processes as they happen, but developments using artificial intelligence are building models and improving analytics, Zornio said.

Still, good data is essential for AI applications in manufacturing, he said.

“You can have a lot of raw data, but if that data isn’t in the right format or in the right context, you can’t link the data from this application set to that application set,” Zornio said. “You can’t run the optimization and build the model across those multiple data sets, especially if you’re trying to cover multiple functions within a plant.”

There are several promising use cases for generative AI in manufacturing, he said. For example, a company could use the technology to develop automation systems specifically for each of its facilities.

“GenAI can automatically configure the system by feeding in the system drawings of the physical devices and their connections and creating an initial configuration of the control system,” Zornio said. “GenAI can do that engineering work for us.”

Generative AI can also be used as a “super assistant” for products, taking in all the documentation, support interactions, logbooks and application notes to create a product chatbot that has comprehensive knowledge and can better answer when you ask how the product works, he said.

“Everyone who makes a product wants to build something like a super product assistant,” Zornio said.

Poor data quality slows down AI projects

The problem has never been that there is enough data in factories and manufacturing plants. What is more important is the ability to access and use this data, says Jon Sobel, co-founder and CEO of Sight Machine, a provider of a data platform for industrial companies.

Data problems are one of the main reasons AI projects in manufacturing have stalled, Sobel said.

The manufacturing industry produces about twice as much data as any other sector of the economy, but so far little of it has been used for productive purposes, he said. One problem is that the data is generated at different times and from different sources that do not interact with each other.

“The real challenge is that most of this data goes unused because of the way we historically develop applications – you build 50 apps and none of them can talk to each other,” Sobel said. “So the paradigm has to change to structure the data, know enough about the data and then start querying it.”

Once manufacturers have properly structured and prepared the data, new AI advances are the ideal technology to make that data usable, he said.

“We’ve been talking about the importance of data for a long time – and nobody cared,” Sobel said. “With AI, people think that if they want to use AI, their data has to be right. AI is a catalytic moment in the conversation about good data.”

Jim O’Donnell is a senior news editor at TechTarget Editorial, covering ERP and other enterprise applications.