AI for Analyzing Shifts in a Conversation
Yair Neuman · 2025
Conversations evolve in a nonlinear way, and this chapter explores this dynamic using large language models (LLMs). I show how to guide the AI in analyzing a conversation by paying close attention to the way it unfolds and picking out transition points in the conversation. This approach provides a structured methodology for identifying shifts in conversation and offering insights into the participants’ underlying power dynamics and evolving roles. Using a tantalizing scene from No Country for Old Men , the chapter demonstrates the application of a sliding window analysis, segmenting conversations into overlapping utterances to examine ambiance, speech acts, and relational dynamics. Applying this approach to the scene, the AI clearly exposes the dynamics and the way it shifts from small talk to a game of life and death.