Skip to main content

Industrial-grade Artificial Intelligence Makes Manufacturing Smarter

When Jim Preston awakes 90 years too early aboard the Starship Avalon in the movie Passengers, his conversation with an AI-enabled hologram seems pretty natural. As a viewer, you don’t really question the scene because you’re accustomed to interacting with the AI-enabled voice assistant, your bank’s chatbot, or the navigation app in your car. It’s technology adoption at its best. You not only accept the new skill, but you also assimilate it into your daily routine. Life is getting easier, and you hardly even notice how it’s happening.

This shouldn’t come as much of a surprise to any of us. The trends are clear in the consumer world. The smart speaker – a technology that just hit the market in 2014 – has been adopted by 94 million people in the U.S. with many households owning more than two. 

In the industrial world, advanced technology adoption is also accelerating. Last year, 67% of companies increased their adoption of automation and AI, according to McKinsey. IDC forecasts that spending on AI technologies will grow to US$97.9 billion by 2023 — more than two-and-a-half times what was spent in 2019. And global GDP will be up to 14% higher in 2030 as a result of the accelerating development and adoption of AI, the equivalent of an additional $15.7 trillion, according to PwC.

AI is coming to all industries, and for the data-rich world of manufacturing, the AI must be industrial grade.

AI is coming to all industries, and for the data-rich world of manufacturing, AI must be industrial grade. The first real applications of machine learning have already found a place in plant-floor activities such as rapid visual inspection, anomaly detection, predictive maintenance, audio diagnostics, automatic sortation, and more. And that’s just for starters.

Brewers can already use AI to do predictive modeling to account for the natural variation of ingredients. Manual processes are replaced by AI, allowing them to achieve consistent taste and quality for every batch of beer. Another use case is in press shops of the automotive industry. By using machine learning algorithms, automakers can reduce scrap parts, increase throughput, and ensure consistent quality.

Things get more interesting when you blend edge and cloud computing. Neural networks can help plant managers read between the lines and see complex connections that aren’t evident to the human eye. This is possible by linking high-resolution edge platform data from the factory floor with the computing muscle of the cloud. As a result, production can become more efficient, flexible, and reliable.

AI is a natural part of the digitalization of manufacturing, which is why open IoT operating systems like MindSphere have become so important. With MindSphere, users can collect and view data gathered from a machine (or a whole fleet of machines) and also use algorithms to generate recommendations for action, such as how a machine can be more efficiently used or how its operation can be optimized.

AI is a natural part of the digitalization of manufacturing, which is why open IoT operating systems like MindSphere have become so important.

As with digitalization, AI’s benefits for manufacturing encompass the entire lifecycle – from product design to production to real-world performance. We’re accustomed to thinking about feedback on product performance. But with digitalization and deep learning, we can also create a “feed-forward” loop – using data from real-world performance so that the next generation of that widget can be faster, lighter, or cheaper. AI can also help in the generative design process by creating millions of models, analyzing them, and narrowing the list to 100 or so variations with the highest probabilities of success.

Deep learning gives manufacturers the opportunity to take the broadest view of their business possible. This involves opening the aperture of the lens from small improvements of existing processes to entirely new products or even business models, partnerships, and ecosystems.

As with any groundbreaking new approach or technology, human behavior and organizational culture need to be part of the conversation. High-tech factories are not going to run themselves. And if the consumer world tells us anything, when AI systems seem more human, they tend to be more successful. That’s why people accept the suggestions of the airline AI system if it pauses briefly to search for flights, creating the impression that it is thinking. Every technology has a learning curve! 

It is easy to understand why AI is predicted to give global GDP such a significant boost by 2030. AI is inspiring manufacturers with new insights about what we make, how we make it, and how we go to market. It is challenging us to prioritize continuous learning as a key part of our organizations’ growth plans. And it is opening entirely new ecosystems of cooperation among companies. I’m optimistic about what AI means for all of us working in manufacturing and the customers we serve. 


Raj Batra

Raj Batra

President of Siemens Digital Industries, USA

Raj Batra is the President of Siemens Digital Industries USA. He is a member of the Executive Committee of the Manufacturers Alliance and the immediate past Chairman of the Board of Governors for the National Electrical Manufacturers Association (NEMA)

Many thanks to Raj for sharing his insights.

Opinions expressed by contributing authors are their own.

For more, see definitions from Siemens

  • Artificial intelligence
    • In its truest sense, artificial intelligence refers to applications in which machines perform tasks that would normally require functions of human intelligence such as learning, judging, and problem-solving. Tools and technical solutions are being developed for this purpose, enabling humans to work better by extending their abilities.
  • Machine learning
    • Machine learning (ML) is what underlies the actual “intelligence” in AI. Computers are trained to recognize patterns in unstructured datasets using algorithms and to make decisions by themselves based on this “knowledge.” The goal is to have the machine learn from the data and, based on this, use the experience it acquires to constantly improve its ability to perform its tasks.
  • Deep learning
    • Deep learning (DL) relies on the use of deep neural networks. The computer accesses data at several node levels simultaneously to identify connections, draw conclusions, and make both predictions and decisions. Self-learning algorithms enable the machine to solve even complex non-linear problems by itself and to interact without instructions.