Through the looking glass: challenges and pitfalls of computational storytelling
Storytelling is a mirror we shine on our own lives, and one which we often look at to try and understand ourselves. However, we know that the correspondence between the world in stories and the real world is far from exact.
Computational storytelling attempts to model the human ability to make stories. In this sense, it should provide a computational model of the relation between humans, their real world, and the stories they make.
Stories are often considered useful or not depending on how faithfully they capture reality. Unbelievable or obviously exaggerated stories tend to be frowned upon by listeners. Yet stories that are imaginative to the point of being original or surprising are often considered better than their non-imaginative counterparts. It seems the storyteller is expected to walk a thin line between straight representation of reality and subtle transgression of veracity employed to spice up material that is otherwise faithful.
The talk will focus on challenges and pitfalls that surround this storytelling task, where both completely faithful rendering of reality and too large a departure from it are likely to lead to poorly valued outcomes.