What is sought in GEO is the ability of a system to augment its knowledge base as it encounters new information while interacting with students. This is a form of knowledge acquisition termed 'learning from being told' and is an active area of investigation within AI. As seen in the dialogue above, GEO must possess a good natural language processor if it is correctly to interpret input from the student in the flexible manner portrayed. It will also need to keep track of what each student knows about geography and of any misconceptions he or she might have. Knowledge of the student, developed over many interactions, will guide the interaction, leading the student to review hazy facts or explore new areas.