Closer Worlds: Using Generative AI to Facilitate Intimate Conversations
Tiffany Chen, Cassandra Lee, Jessica R Mindel, Neska Elhaouij, Rosalind W. Picard
Abstract
Deep emotional intimacy is a foundational aspect of strong relationships, but the digital tools we use to communicate often limit rather than empower our feelings of connection. Two compelling strategies technologists have used to counteract such trends include games and generative AI art. In this paper, we design and test Closer Worlds, an ML-assisted 2-person game that fosters emotionally intimate conversations through co-creative world-building. We explore design principles inspired by facilitation methods and assess their effectiveness in a pilot study. We find that Closer Worlds elicits some self-disclosure behavior, but less than a social game without generative AI. However, participants clearly enjoy the unique affordances offered by visualizing shared values, which suggests that this method offers a comfortable and novel avenue for meaningful conversations. We conclude by discussing future ways in which co-creative games might leverage generative techniques to foster pro-social environments.