The automaker produces thousands of parts every day, yet it doesn’t learn about its own production until up to two days later. The data exists, but it’s scattered across dozens of systems, in manually maintained spreadsheets, and in the minds of experienced foremen. Jana Vydrová, who works at Škoda Auto on digital component production management and shop floor IT, explains in an interview why you shouldn’t start with AI when it comes to AI and what’s really the hardest part of digitization. On Tuesday, October 6, she’ll speak on this topic at the Technical University of Liberec as part of the University Doctoral School.
Jana Vydrová will visit TUL to deliver a lecture titled “The Pitfalls of Digitalization in the (Not Only) Automotive World.” CXI TUL has been collaborating with Škoda Auto for a long time on the digitalization of production, including work on the mVIS visualization platform. As a department at a technical university, it emphasizes links to the real world and conducts research in collaboration with industry. That’s why, before her lecture, we asked Jana what industry can gain from research, where to start with AI, and what is truly the most difficult aspect of digitization.
Where do you see the greatest potential for collaboration between industry and universities and research organizations?
Industry has ideas and, unlike universities, financial resources. However, it lacks flexibility and the ability to quickly test hypotheses in a lab or test environment. Most ideas are discarded before they can be tested, and that’s exactly where I see room for collaboration. At universities
and especially in research organizations—as the name implies—there are plenty of curious people who are motivated by a constant drive to improve things, make discoveries, and learn new things.
What do you think a research environment can offer a large company that the company would find difficult to create on its own?
I’d like to build on my previous answer. The atmosphere in which changes take place is definitely worth mentioning. In a large company, changes are usually unwelcome and perceived as a disruption to established rules. To put it a bit dramatically, the motto is “if it ain’t broke, don’t fix it.” A research institution, on the other hand, would likely be very unsuccessful without change.
What would you recommend to companies that want to start with digitization, AI, or process modernization but don’t know where to begin?
It’s definitely important to understand what the company or team wants to achieve. If the goal is simply to implement some AI use case, adoption usually doesn’t go well. You need to address real problems that people are facing. Whether it’s customers or internal employees, the motivation is the same: if a tool doesn’t solve my problem, I’m not going to use it. And it should be a real problem, not one that’s just in management’s heads.
Another essential prerequisite is having clean data, including context and metadata. Before anyone embarks on deploying AI, they should work on ensuring data availability and making it possible to collect it automatically, without manual entry.
What’s the hardest part of digitizing manufacturing today—technology, data, or changing processes and the way people work?
I ask myself this question every day.😊 I don’t want to do anyone an injustice, so I’m basing my answer mainly on my own observations. For me, the hardest part is changing my own habits and starting to think differently. This is related to processes, established routines, people’s work, and a change in mindset in general. Data is, of course, also very important, but a lot can be solved through technology. If we didn’t bend technology just to avoid having to change our habits, digitization would certainly progress faster.