Predicting the physical condition of documents: Dalgocol, a research program in artificial intelligence at the BnF. Interview with Philippe Vallas • DALGOCOL (“Data mining and ALGOrithms predicting the state of COLlections”) was launched in 2018 by Alaa Zreik as part of his PhD thesis. The goal is not to come up with an operational tool, but to test the possibility of using artificial intelligence methods on the computer data produced by the Bibliothèque nationale de France (BnF) as part of its conservation activities. Why use AI? One of the challenges is to take advantage of very large amounts of data: for example, about two million documents were tested at the beginning of the work. The automation makes it possible to cross-reference different types of data to help specialists decide whether or not to remove a resource from library access while it is being restored. While the work is led by Alaa Zreik, it solicits the skills of BnF professionals. Philippe Vallas is currently assistant to the director of the Conservation Department at BnF, and is responsible for coordinating physical conservation actions and measures. For the “Bulletin des bibliothèques de France”, he looks back on the opportunities that have arisen thanks to Dalgocol.