Uncovering Value Dynamics in Conversation: A New Role for AI in Joint Value Creation

In collaborative and high-stakes environments, decisions are rarely made on facts or data alone. They are shaped by deeply held, often unacknowledged values. The core challenge addressed in this thesis was to understand the values that influence conversations in decision-making settings. The research focused on value misalignment, the illusion of shared values, and the tensions or opportunities that often remain unspoken in group dynamics.

To explore this, Vidhi Chopra applied Natural Language Processing (NLP) as both a method and a provocation. Using a Research through Design approach, she developed a low-fidelity prototype that analyzes conversation transcripts to detect implicit value cues. These cues were not found only in direct statements but also in the framing of ideas, metaphors, and even what remained unsaid.