Ms. Maduekwe, why is ARC now tackling the topic of AI with a study - isn't AI already firmly established in the industry?

Ebele Maduekwe: Yes and no! The topic of artificial intelligence has been around for many years, but it is only recently that we have seen an increased interest in AI in industrial applications. We can identify three main reasons for this delayed interest: AI is currently still experiencing very rapid innovation, the provider environment is still constantly changing and the potential of AI applications is still difficult for users to assess. However, we see a clearly defined added value in the use of AI in the areas of quality control, machine performance management and robotics.

The term machine learning is often used for AI. How do you differentiate here?

Ebele Maduekwe: We define AI in a manufacturing environment as a system that perceives its environment and takes action to maximize its chances of achieving its goal - for example, a robot moving from A to B in an unstructured, changing environment. Machine learning, on the other hand, comprises various techniques that enable a machine to learn about its environment in order to achieve predefined goals. In other words, machine learning is a tool for AI.

What have you found out? How do you 'buy' AI? Is it a software, a hardware module or a licensed algorithm that you buy?

Ebele Maduekwe: Artificially intelligent systems and devices come in various forms: As hardware with a dedicated AI chip, as software with AI libraries that you can use to create your own AI model, or with ready-made AI models as plug & play modules. With new applications and use cases, it is also possible to use new constructs of hardware, software and services as a stand-alone solution or as a combination. - This diversity is what makes it so difficult for users to form their own opinion of the right solution.

What skill sets are required to implement AI?

Ebele Maduekwe: Depending on the industry, the required expertise ranges from zero IT knowledge to specific automation know-how. For simple use cases, only data science knowledge may be required, but for complex processes, knowledge from the fields of automation and IT. Of course, several specialists can come together to solve a problem. So there is no one-size-fits-all recipe.

In which sectors do you see the 'first movers'? In which applications?

Ebele Maduekwe : Automotive production is currently benefiting the most from AI. At the moment, we see the highest acceptance in the automotive and robotics sectors. In terms of applications, we are seeing the highest acceptance in image processing for quality control. In image processing systems, products and solutions have been developed in which quality control is analyzed in real time using simple artificial neural networks through to complex neural networks with deep learning. These solutions are already being used in the electronics and semiconductor industries and are also being rolled out in sectors such as wood processing.

How do you see the market developing over the next decade?

Ebele Maduekwe: We see steady growth as more and more manufacturing companies learn to use AI and recognize its added value in their operations. After machine vision in the automotive industry, the other industries and application areas will catch up. We also expect new products to be developed for both common use cases and specific manufacturing use cases. And most importantly, the development of more cost-effective solutions will drive the adoption of many AI technologies and applications.