Motius interview: "There must be a clear use case"

Handelsblatt Research Institute

Mr Bahrouni, which digitalization topics are currently the focus of production companies?

Zied Bahrouni

For a while, the main focus was on digitally mapping manual processes. Now, the focus is more on bringing digital processes closer to the physical world. In terms of production, this means merging the digital processes at headquarters with the physical production chain and coordinating the two.

For example, the aim is to improve production control. This requires algorithms, sensors, actuators and robotics so that the digital and physical worlds can interact with each other. In this respect, the question is no longer what can I digitize, but how can I improve the physical world - for example, production - with new technologies.

Handelsblatt Research Institute

Which trend technologies will the focus be on in the coming years?

Zied Bahrouni

This is difficult to narrow down. Generally speaking, it's about technologies that increase productivity or are used to develop new products. In my opinion, the trial and error phase is largely over. Today, there has to be a clear use case for the use of technologies.

Handelsblatt Research Institute

Do you see such a use case in the metaverse?

Zied Bahrouni

Not yet, at least not in the necessary size and breadth. There has been a lot of talk about this technology at conferences, but the use case is still missing.

Handelsblatt Research Institute

How do you assess the potential of quantum computing?

Zied Bahrouni

Huge, but not tomorrow. Quantum computing will radically change many industries. But the technology is not yet mature enough. But it will come.

Handelsblatt Research Institute

What potential do you see in AI? Do intelligent applications increase productivity or enable the development of new products?

Zied Bahrouni

There is definitely great potential for increasing productivity. When we talk about AI, we are currently mainly talking about large language models. The advantage here is that you can communicate with the models in natural language - in any language. For example, new employees on the store floor of a manufacturing company can then be trained more quickly, as everyone can carry out the training themselves by communicating with the models. And the language doesn't matter. This reduces the time required for onboarding.

Another use case that is becoming increasingly important is software development. LLMs create parts of the software automatically, which is of great value and increases productivity.

With regard to products, LLMs also enable a new interface - in cars, for example - with customers. They can interact with the system in natural language.

Handelsblatt Research Institute

How long does it take for AI applications to pay for themselves?

Zied Bahrouni

Especially when implementing LLM technologies, which are relatively simple on the application side, the ROI is usually faster than with other technologies. And they are also less expensive in comparison.

Handelsblatt Research Institute

Which factors are decisive for the success of AI applications?

Zied Bahrouni

A key success factor in AI projects is the transfer of technology to specific use cases. The challenge in many companies is that the departments with a focus on technology - such as the research or innovation department - often work in isolation from the business departments. This shows that it is partly structural and methodological aspects that significantly increase the chances of success of an AI project, and not necessarily the technology itself.

AI projects should not be set too small. The benefits should be so great that they can be perceived throughout the entire company. This also makes larger budgets for further projects likely. A use case in just one niche does not achieve this.

Handelsblatt Research Institute

What role do the AI skills of the workforce play?

Zied Bahrouni

Companies do not need to develop all AI skills internally. However, they do need a basic skill set so that companies are not completely dependent on external partners.

Handelsblatt Research Institute

A few years ago, the lack of AI talent was still a major challenge for companies. Are they now less reliant on talent for AI applications?

Zied Bahrouni

Yes, the development of AI applications based on AI models is generally easier today than traditional software development. Foundational models provide the basis so that the models themselves no longer need to be developed, only the applications based on them. This also makes the recruitment of AI talent less of a bottleneck for successful implementation in companies.

Handelsblatt Research Institute

Thank you very much for the interview, Mr. Bahrouni!