Perspectives on artificial intelligence in production

(L-R) Sven Krause (moderator), Prof. Michael Zäh, Sebastian Heinz

By: Kathryn Gerardino-Elagio

During the EMO 2025 Preview in Frankfurt am Main on July 9, 2025, International Metalworking News for Asia joined over 100 invited participants from around the world to explore key trends shaping the future of manufacturing. One of the highlight sessions was the panel discussion, “Perspectives on Artificial Intelligence in Production: How Will We Surprise Our Customers in the Next Five Years?,” featuring Sebastian Heinz, Founder and CEO of statworx (Frankfurt am Main), and Prof. Michael Zäh, Chair of Machine Tools and Manufacturing Technology at the Technical University of Munich.

AI in Manufacturing: Not a Newcomer, but a Game Changer

Heinz opened the conversation by noting that AI, machine learning, and statistical data analysis are far from new in manufacturing. Decades-old techniques such as Statistical Process Control (SPC) have long been used to analyse production data and optimise output. What’s new, he emphasised, is the pace of innovation and the maturity of technology on the production floor, which may be ahead of other departments in unlocking AI’s value.

Zäh agreed, pointing out that while AI’s potential spans every level of an enterprise—from shop-floor processes to management decisions—the greatest untapped opportunities remain in small and medium-sized enterprises (SMEs). Larger manufacturers have already adopted AI in targeted ways, but SMEs often face barriers such as limited financial resources, lack of awareness, and the absence of “plug-and-play” solutions.

Cultural Differences in AI Adoption

Heinz observed that openness to AI varies greatly by region. In his experience, U.S. manufacturers—and those in parts of Asia—tend to embrace new technologies more readily than their German counterparts. This, he argued, reflects broader cultural attitudes toward innovation and collaboration with smaller, more agile companies.

While Germany’s manufacturing base has strong technical expertise, Heinz believes a mindset shift is needed to seize AI’s full potential. Without timely investment, he warned, European manufacturers risk falling behind global competitors who are already embedding AI across operations.

Practical Applications: From Welding to Scheduling

To ground the discussion, Zäh offered three practical examples from his research:

1. AI-based welding inspection – Using camera systems to detect cracks, pores, and surface defects in friction stir weld seams, replacing manual inspection for large components such as rocket tanks and aircraft parts.
2. Predictive maintenance for machine tools – Replacing spindles, linear drives, and other components based on real-time condition monitoring, rather than fixed schedules, aligning maintenance with production plans.
3. AI-driven line scheduling – A genetic algorithm-based system that can optimise assembly line schedules in 10 minutes, a task that previously consumed two hours of a supervisor’s day.

Zäh stressed that these solutions already exist, either from research labs or specialised service providers, but remain underutilised due to a lack of awareness or integration capability.

Competitive Advantage: Build, Don’t Just Buy

Both panelists agreed that buying off-the-shelf AI tools may boost productivity but will not deliver lasting competitive advantage—especially once those tools become widely available. Heinz urged companies to build internal capabilities for applying AI to their unique data and processes. This, he said, is where the real differentiators will emerge in the next five to ten years.

“AI’s promise isn’t just cutting costs,” Heinz said. “It’s about creating something your competitors can’t easily copy.”

Education and Knowledge Gaps

The conversation also turned to education. Zäh described how the Technical University of Munich has been teaching AI fundamentals—such as neural networks and genetic algorithms—alongside applied scenarios in areas like autonomous vehicles and predictive maintenance. However, he cautioned against introducing AI too early in formal education, advocating instead for strong foundations in mathematics, languages, and critical thinking.

For industry professionals, he encouraged direct collaboration with universities, noting that networks of manufacturing researchers exist in Germany, Canada, Japan, and elsewhere. “Get in contact with science,” Zäh urged. “Develop something together.”

Speed of Adoption and Infrastructure Needs

While media headlines highlight “dark factories” and fully AI-driven production lines, Heinz warned that such examples can be more marketing than reality. Many facilities still lack machines capable of generating the high-quality, contextual data that AI systems require. Upgrading infrastructure to produce actionable data, he argued, is the necessary first step toward realising AI’s full promise.

Both speakers also highlighted the interoperability challenge: without common data standards or interfaces, integrating AI into diverse production systems remains a hurdle—especially for SMEs.

The Next Five Years: Innovation by AI

Looking ahead, Heinz predicted that within five years AI systems will be capable of generating entirely new ideas, driving innovation not just in manufacturing processes but across sectors from medicine to materials science. This capability could trigger an acceleration in research and development unlike anything seen before.

Zäh took a more grounded view, foreseeing tangible results in predictive maintenance, quality assurance, and productivity gains. While visions of empty, fully automated factories have circulated since the 1970s, he expects human work to remain essential, albeit in evolving forms.

Impact on Customers and Markets

From the customer’s perspective, the widespread adoption of AI could mean higher-quality products, faster production, and, eventually, lower prices—though only if cost savings are passed down the value chain. Heinz suggested that a bifurcation could emerge between fully AI-produced goods and those valued for human craftsmanship, with both coexisting in the market.

Drawing a parallel to agriculture’s century-long productivity gains, Zäh predicted a similar trajectory for industrial manufacturing: fewer workers producing more, with the challenge being to identify where human skills remain irreplaceable.

Conclusion: Act Now or Fall Behind

The panel closed with a clear message: adopting AI in production is not optional. Companies that delay risk losing competitiveness to early adopters who develop AI capabilities tailored to their operations. Whether through collaboration with research institutions, investment in infrastructure, or building internal expertise, manufacturers must act now to capture AI’s transformative potential.

As Heinz put it, “We will always need to work—it’s human nature. The question is, what will that work look like in an AI-driven future?”