Breaking Black-box AI

Explore the limitations and failure cases of popular machine learning algorithms in this conference talk from GOTO Berlin 2019. Dive into live demonstrations showcasing unexpected outcomes from linear regression to cutting-edge deep learning techniques. Gain insights into practical examples where treating AI as a black box can lead to unintended consequences. Discover the importance of understanding algorithm limitations in various fields, including social problems, medicine, and data science. Learn why critical thinking and domain expertise remain crucial when implementing AI solutions, even as machine learning becomes more accessible through APIs and pre-built models.