Explore the intersection of machine learning and sequential experimental design in this hour-long lecture from the Simons Institute. Delve into the challenge of quantifying the value of information in experimental planning, traditionally analyzed using probabilistic models like Gaussian processes. Discover how machine learning offers improved flexibility and representational power, while also introducing new challenges in algorithm design. Survey multiple projects examining this challenge from various perspectives, including frequentist ensembles, Bayesian deep kernel learning, and direct modeling of information value using deep neural networks. Gain insights into the potential of machine learning to enhance sequential experimental design and its implications for future research in this field.