E Contextibus Unum: When Should Your AI Share Your Information with Others?
Abstract
As large language models (LLMs) become increasingly integrated into collaborative environments, questions arise about how user contexts such as prompts, conversation histories, and preferences might be shared across users to enhance collective intelligence and personalization. Although technical implementations of context sharing have been explored, it remains unclear how to design interfaces that appropriately engage users with context sharing. This study investigates how users perceive the trade-offs between usefulness, privacy, and control when their contexts are shared with other users through LLM-mediated interactions. Through a Wizard-of-Oz study with scenario-based probes, we examine both the sharer’s perspective (those whose context is being used) and the receiver’s perspective (those benefiting from others’ contexts). Our work aims to surface guidelines for designing LLM interfaces for transparent, user-centered context sharing systems that balance collective benefit with individual privacy and autonomy. This provides engineers with opportunities to design LLM assistants that leverage aggregated user data to provide useful insights. Our key findings reveal that: (1) context sharing acceptability varies by domain: professional contexts showed high acceptance while health contexts demanded expert authority over community insights, and mental health contexts produced polarized responses based on personal disclosure comfort; (2) most participants preferred statistical aggregations over individual anecdotes, demonstrating an LLM-specific trust mechanism where data-driven validation matters more than personal stories; and (3) control is essential for acceptance, but must avoid decision fatigue through domain-based preferences and keyword rules rather than requiring per-interaction consent decisions.
Keywords: Large Language Models, Context Sharing, Privacy, User Control, Human-AI Interaction, Wizard-of-Oz Study
DOI: 10.54941/ahfe1008262
Cite this paper
More from this volume
- AI-Enhanced Live-Streamed English Language Learning: Designing and Evaluating an Intelligent Platform for Chinese Learners
- "They Will Move": Confirmation Bias, Expectation Frames, and Premature Monitoring Drop Off in COLREG Encounters
- Automation transparency communication does not improve supervisory control performance, but shapes operator trust
- The influence of neuroticism on startle and surprise response: a flight simulator study
- EEG-Based Classification of Reward, Penalty, and Movement Events Across Four Video Games of Varying Complexity
- The Effects of Task Similarity and Time Pressure on Task Performance Following Interruptions in Complex Human-Computer Interaction Systems
- Trusting the Chain, Not the Black Box: A Provenance-Based Architecture for Human-Machine Teaming in Military Autonomy
- Design of a Simplified Humanoid Robot Face: Cultural Differences and Emotional Synchronization
- From Visual Observation to Auditory Interaction: A Human-Centered Tangible Music System Based on Physical Objects
- The Impact of Scarcity-Based Dark Patterns on User Performance, Perceived Darkness, and Workload in Subscription Services
- A Behavioral Observation Tool for Diagnosing Human–AI Interaction Profiles in Engineering Design Education
- The Mediated Moon: Semantic Mediation Between Gesture and Generative Image


AHFE Open Access