The way we search for information is undergoing a radical transformation thanks to generative AI. Large language models (LLMs) are changing the game, but they also present new challenges for researchers. A recent workshop by the Computing Community Consortium (CCC) brought together experts to explore the future of information retrieval (IR) in this new era. They identified key areas where research is crucial, including developing better ways to evaluate these new AI-powered search systems, training them more effectively with human feedback, understanding how users interact with them, and addressing the societal implications of this technology. One exciting area is the development of “digital twins” – personalized AI assistants that understand individual user preferences and contexts to provide tailored information. Researchers also emphasized the need for more efficient models that consume less energy and are accessible to a wider range of users. Another focus is on creating AI agents that can proactively gather information and perform tasks on our behalf, potentially leading to revolutionary changes in how we interact with the digital world. Finally, the workshop discussed the long-term vision of creating foundational models specifically designed for information access and discovery, promising to enhance human intelligence in unprecedented ways.