
Dr. INDERJEET MANI
Monday, July 20th, 2026
1:00 – 2:00pm
SPRK 229 (Formerly Global)
Abstract
Story understanding and storytelling have been longstanding problems in AI. Using NarrativeML, a structured representation of narrative elements and relations, I examine how LLMs recover narrative structure, infer causality and motivation, and generate stories from formal representations. The results reveal substantial narrative capabilities, including the modeling of perspective, temporal order, character goals, and plot structure. However, there are persistent limitations in reasoning across long distances in text, as well as a lack of embodied understanding and cultural knowledge. This assessment sheds light on the nature of narrative intelligence, and also carries broader sociocultural implications for generative AI.
Biography
Hosted by IAI NLP Group