Enabling Partial Participation in Remote Meetings
Authors
Paper Title
Enabling Partial Participation in Remote Meetings
Publication Info
- Topic area: Remote collaboration and AI-mediated meeting participation
- Keywords: Partial participation, remote meetings, AI proxies, multitasking, LLM updates, social presence, mental workload, meeting comprehension, hybrid work, ProxyMe
Background and Problem
- Problem / challenge: Current collaboration platforms treat participation as a binary state (fully present or absent), failing to accommodate intermittent or multitasked engagement in remote meetings.
- Significance: Addressing this gap is critical as remote and hybrid work become more prevalent, with workers often multitasking or attending overlapping meetings.
- Motivation and related work: Prior research highlights the limitations of existing meeting tools in supporting flexible participation, the challenges of multitasking, and the potential of AI-generated updates and conversational agents. However, these approaches have not been fully explored for enabling meaningful contributions from absentees during live meetings.
Solution
- Proposed approach: ProxyMe, a prototype system that enables partial participation in meetings through real-time updates and AI-mediated decision deferrals.
- Novelty:
- Conceptualization of partial participation as a middle ground between synchronous and asynchronous meeting involvement.
- Development of ProxyMe, which combines real-time updates and agent-mediated deferrals to support multitasked participation.
- Exploration of how update frequency and response styles influence social presence, mental workload, and comprehension in multitasking scenarios.
- Procedure and key techniques:
- Real-time meeting updates generated by an LLM (Google Gemini 2.5 Flash-Lite) in two styles: Low-Detail/High-Frequency (LD/HF) and High-Detail/Low-Frequency (HD/LF).
- Agent-mediated deferrals prompting user responses via multiple-choice (MC) or text-input (TI) styles.
- A mobile-friendly interface for monitoring updates and responding to deferrals.
- A simulated meeting environment for controlled user study.
Results
- Concrete findings:
- Social presence ratings averaged 3.53/7, indicating moderate connection with meetings.
- Video comprehension scores were slightly lower with LD/HF updates (-1.03 effect size), while meeting comprehension showed minimal differences across conditions.
- Mental workload was moderate to high (mean RSME = 79.79/150), with no significant differences between conditions.
- Participants preferred MC response style over TI, regardless of update style.
- Advantage over baselines:
- ProxyMe enables real-time engagement and decision-making, reducing reliance on post-meeting summaries and enhancing inclusion in ongoing discussions.
- Updates and deferrals fostered a sense of presence and allowed participants to contribute semi-synchronously.
- Experiments / evaluation:
- A 2 × 2 factorial user study with 24 participants (aged 19–30) evaluated the effects of update style (LD/HF vs. HD/LF) and response style (MC vs. TI) on comprehension, social presence, and mental workload.
- Participants multitasked by watching a video lecture while engaging with simulated meetings through ProxyMe.
- Data collection included quizzes, questionnaires (e.g., Networked Minds Scale, RSME), and structured interviews.
- Limitations and future work:
- Small, student-based sample limits generalizability to professional settings.
- Simulated meetings lacked the complexity and social dynamics of real-world scenarios.
- Future work should explore live meeting deployments, richer engagement methods, and adaptive system configurations.
Summary
This paper introduces partial participation as a novel paradigm for remote meetings, addressing the limitations of binary participation models. The ProxyMe prototype demonstrates how AI-generated updates and agent-mediated deferrals can enable multitasked engagement, fostering moderate social presence and comprehension. A user study revealed trade-offs between update frequency, response styles, and mental workload, highlighting the contextual and individual variability in user preferences. While ProxyMe offers a promising approach to flexible meeting participation, future research should focus on real-world applications, richer interaction methods, and balancing productivity with well-being.
Research Questions / Practical Problems
Question signals indexed for this paper.
- 83%
LADICA: A Large Shared Display Interface for Generative AI Cognitive Assistance in Co-located Team Collaboration
CHI '25· Human-LLM Collaboration +1
- 71%
Cognitive Bridge: AI-Generated Boundary Objects for Cross-Functional Collaboration
CHI '26· Human-LLM Collaboration +2
- 67%
An Evaluation of Situational Autonomy for Human-AI Collaboration in a Shared Workspace Setting
CHI '24· Human-LLM Collaboration +1
Based on Jaccard similarity of research subtopics & professions (≥60%)