All news at TU Wien

Article in the Industry Magazine

Data Projects: 60 Percent of Success Lies in Preparation Technology is rarely the problem. Data-driven projects usually fail because they are too complex, too abstract, and not structured well enough. Here is what those responsible can do to prevent such failures.

AI and Industrial Data Science graphic

by Ansari & Steiner

“Would you consider it reasonable to travel from one district of Vienna to another by airplane?” asks Fazel Ansari, before answering his own question: “Of course not.” Yet that’s exactly what many companies are currently doing when they think about using AI systems, says the university professor for Data-Driven Maintenance Management and head of the Research Unit for Production and Maintenance Management at TU Wien. They invest in complex technologies where simple solutions would suffice, or worse, they fail because they start from the wrong foundation.

Andreas Steiner, a research associate at the Research Unit for Production and Maintenance Management at TU Wien, offers another example that highlights the risks: “A visual defect detection tool with a 95 percent accuracy rate sounds impressive. But when inspecting aircraft turbines, you might want something more precise. Insufficient accuracy in this context can have fatal consequences and completely negate the system’s overall benefit.” Read more …, opens an external URL in a new window