Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Caught in a Mafia Romance: How Users Explore Intimate Roleplay and Narrative Exploration with Chatbots

Julia Kieserman, Cat Mai, Sara Lignell, Lucy Qin, Athanasios Andreou, Damon McCoy, Rosanna Bellini

TL;DR

It is found that users problematize both the excessive and insufficient sexualized content in such interactions which warrants novel digital-safety features, and that users want to engage in intimate role-play with young adult, masculine-presenting characters that place users in a position of inferior power in well-defined scenarios.

Abstract

AI chatbots, built using large language models, are increasingly integrated into society and mimic the patterns of human text exchanges. While previous research has raised concerns that humans may form romantic attachment to chatbots, the range of AI-mediated interactions that people wish to create for themselves or others with chatbots remains poorly understood, particularly given the fast evolving landscape of chatbots. We provide an empirical study of Character.AI (cAI), a popular chatbot platform that enables users to design and share character-based bots, and synthesize this with an analysis of Reddit posts from cAI users. Contrary to popular narratives, we identify that users want to: (1) engage in intimate role-play with young adult, masculine-presenting characters that place users in a position of inferior power in well-defined scenarios and (2) immerse themselves in boundless, fantasy settings. We further find that users problematize both the excessive and insufficient sexualized content in such interactions which warrants novel digital-safety features.

Caught in a Mafia Romance: How Users Explore Intimate Roleplay and Narrative Exploration with Chatbots

TL;DR

It is found that users problematize both the excessive and insufficient sexualized content in such interactions which warrants novel digital-safety features, and that users want to engage in intimate role-play with young adult, masculine-presenting characters that place users in a position of inferior power in well-defined scenarios.

Abstract

AI chatbots, built using large language models, are increasingly integrated into society and mimic the patterns of human text exchanges. While previous research has raised concerns that humans may form romantic attachment to chatbots, the range of AI-mediated interactions that people wish to create for themselves or others with chatbots remains poorly understood, particularly given the fast evolving landscape of chatbots. We provide an empirical study of Character.AI (cAI), a popular chatbot platform that enables users to design and share character-based bots, and synthesize this with an analysis of Reddit posts from cAI users. Contrary to popular narratives, we identify that users want to: (1) engage in intimate role-play with young adult, masculine-presenting characters that place users in a position of inferior power in well-defined scenarios and (2) immerse themselves in boundless, fantasy settings. We further find that users problematize both the excessive and insufficient sexualized content in such interactions which warrants novel digital-safety features.
Paper Structure (41 sections, 13 figures, 7 tables)

This paper contains 41 sections, 13 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Methods
  4. Data Collection
  5. Collecting cAI Descriptions
  6. Sourcing cAI SubReddits
  7. Qualitative Analysis
  8. Study Limitations and Ethical Considerations
  9. Findings Overview
  10. How users interact with chatbots.
  11. Use Case #1: Intimacy and Sexual Roleplay
  12. Desire for Intimacy
  13. Intimate cAI Descriptions
  14. Use Case #2: Narrative Exploration
  15. Story Creation and Fanfiction
  16. ...and 26 more sections

Figures (13)

  • Figure 1: User interfaces depicting chat interactions with cAI chatbots, exemplifying the design of character-based conversational systems. (a) Depicts a bot, equipped with an icon, tagline, and username (author). Users are then presented with an unprompted, pre-authored greeting. (b) Associated chatbot profile, displaying username (author), number of interactions, description, and AI-generated chat starters. Clicking on a chat starter will lead to (a).
  • Figure 2: Process depicting the character-focused chatbot creation system on cAI. A user specifies a character profile via the creation interface through text-based entry, which is then entered into an LLM. The LLM instantiates a character-specific chatbot (broken line box) that can be found by other end users through cAI's search functionality, or via direct link.
  • Figure 3: Our bespoke cAI data collection pipeline; (1) first we identify a site-map of 840-161 chatbot URLs; (2) then, we scraped the 'Character Overview' pages from 784,137 chatbots (CURATED); (3) then we also scraped the pages of the curated list of chatbot creators (EXTENDED); (4) we then merged these into a final dataset of 5.7M chatbots (SNOWBALLED); (5) finally, we selectively sampled for the most popular (POPULAR) and random (GENERAL) samples for further analysis
  • Figure 4: Empirical Cumulative Distribution Function of interactions for chatbots in CURATED, chatbots that are part of EXTENDED but not of CURATED, and their combination in SNOWBALLED.
  • Figure 5: Example of a "Relationship" Chatbot that potentially suggests intimacy.
  • ...and 8 more figures