Explore the crucial yet often overlooked aspect of choosing prior distributions in statistical modeling and Bayesian machine learning. Delve into the challenges practitioners face when specifying priors that accurately reflect their beliefs, and discover innovative approaches to selecting priors without relying on computationally expensive cross-validation procedures. Learn about prior elicitation techniques for transforming tacit human knowledge into valid distributions, and examine an automated method for choosing priors in Bayesian machine learning. Led by Professor Arto Klami, an expert in probabilistic inference from the University of Helsinki, this 48-minute talk provides valuable insights for researchers and practitioners seeking to improve their use of priors in statistical and machine learning models.