The problem-solving queen and her invisible assistant: data as a fuel for artificial intelligence

With increasing digitalization comes a rise in the volume of data generated in companies – to an almost unimaginable extent. How are we going to handle all this? Lydia Nemec is faced with this question every day. She and her teams process these huge data sets and extract the key information from them to help employees, customers and consumers solve very specific problems. And what is her most important assistant here? Artificial intelligence.

“I realized these computers can help me. But I had no idea what I was doing. I was still so little, and nothing worked,” says the AI expert, adding: “By the time I got the system to the point where it could complete all multiplication tasks, I had long learned my times tables on my own – I no longer needed the system.”

Today, 30 years later, Nemec operates at an interface between data, software and AI in her role as Head of the AI Accelerator at ZEISS, where as she says, “The problems are slightly more relevant here.” For example, it is a matter of optimizing products or even developing new ones, identifying errors in processes or even improving workflows involving communicating with customers.

But even the problem-solvers need help – and they get it from an invisible source: the data set generated in the company. Ultimately, the more progress digital transformation makes, the more data is amassed in the organizations. But this alone is no help to anyone. This data set needs to be translated into information. Only then can the data provide the company with helpful insights.

At ZEISS, Lydia Nemec and her teams rely on another assistant here: “Artificial intelligence is very good at efficiently extracting information. Through machine learning, for example, which helps when processing workflows. Or through deep learning, which, put simply, helps the computer to learn independently through examples,” explains Nemec. “Gathering and storing data does nothing but incur costs. It only becomes useful once information can be extracted from it.”

In her day-to-day work at ZEISS, she encounters more and more moments like the one during her studies. Through her work, she has developed systems that help her to efficiently solve these problems – thanks to data and AI. For example, a modern manufacturing facility in Industry 4.0 can now deliver a huge quantity of data. If trained accordingly, an algorithm could be used to detect when a machine runs the risk of breaking down or which component does not reach the required quality level.

What's more, devices are being made, particularly at ZEISS, which generate incredibly large volumes of data: imaging devices such as microscopes, for example, or measuring machines and medical technology. “A lot of different topics can be addressed with this data,” says Nemec: “For example, we can find out how these machines typically work or how customers actually use our products.” This information is later helpful, for example when improving devices during the product development phase. The Service team can also use it to schedule maintenance or repair works.