Flying commercial aircraft requires the crew’s prediction ca-pability to anticipate the future states of the world (failures, weather situation. . . ), especially those that may harm the safety of the aircraft and passengers. Incident and accident analyses and experimental studies have shown that the occurrence of a conflict during flights (crew/system conflict, pilot/copilot conflict) disrupts the crew’s cognitive capabilities. Therefore the idea is to design a prediction assistant, based on a joint estimation of the state of the aircraft and of the crew’s actions, especially on the autopilot. The prediction of possible conflicting states will allow to enhance the crew’s capability to anticipate the aircraft behavior.