Why AI models often fail in practice
Artificial intelligence (AI) based on machine learning offers opportunities for the life sciences. However, problems often arise in practice. One cause is data leakage, the illicit spillover of information from the training to the test data. Researchers from the Technical University of Munich (TUM), the University of Applied Sciences Weihenstephan-Triesdorf (HSWT), and other research institutions are now advocating for more interdisciplinary collaboration in a new guideline. In this interview, Dominik Grimm, Professor of Bioinformatics, and Markus List, Professor of Data Science in Systems Biology, explain why it is crucial to address this issue now.