← CHI 2026 map

CHI '26 · Honorable mention · full-paper review · confidence medium-high

CHOIR: A Chatbot-mediated Organizational Memory Leveraging Communication in University Research Labs

Sangwook Lee , Adnan Abbas , Yan Chen , Young-Ho Kim , Sang Won Lee

DOI PDF Program page Page JSON

CHOIR is best read as a field-tested workflow contribution rather than a breakthrough algorithm. Its value lies in showing how an LLM chatbot can sit inside lab communication practices to support memory work, while also revealing a real privacy/visibility tradeoff that designers of organizational memory tools often underappreciate.

Axes Lens

Rare contribution shape, typical evidence profile. The point here is not a score. It is to show what kind of claim the paper makes, and whether the evidence pattern is unusual or baseline in this 268 -review set.

Review Summary

CHOIR’s strongest contribution is not a novel model or a benchmark result, but a carefully situated socio-technical system for organizational memory in university research labs. The paper combines document-grounded Q&A, sharing of Q&A for follow-up discussion, extraction of knowledge from Slack conversations, and AI-assisted document updates into one workflow. That integration matters because it addresses the full lifecycle of lab memory work: asking, discussing, capturing, and maintaining knowledge. The field deployment evidence is credible for that claim: four labs, one month, 21 participants, 107 questions, and 38 document updates provide enough activity to observe how the system fits into everyday practice. The most interesting empirical finding is the privacy-awareness tension. Rather than making gaps in documentation automatically visible to managers, private questioning can conceal those gaps, which complicates the assumption that better access necessarily yields better organizational awareness. The paper also surfaces a second, more subtle limitation of contribution: students may hesitate to turn personal experience into universal documentation, which is a practical barrier to knowledge capture in context-rich settings. At the same time, the study should be read cautiously. The deployment is short, small, and context-specific; it lacks a controlled baseline and quantitative performance measures; and one lab being led by the corresponding author raises the possibility of bias. So the paper’s value is primarily descriptive and design-oriented: it offers a plausible system architecture and a set of field-observed tensions that can inform future organizational memory tools, especially in chat-heavy research environments. It does not establish causal superiority over existing lab practices, but it does provide a strong, grounded account of what happens when LLM support is embedded into real lab communication.

What Changed

Canon before

Prior CHI work on organizational memory and lab knowledge management typically emphasizes documentation repositories, search, or explicit knowledge capture; this paper shifts the focus to chatbot-mediated workflows embedded in everyday Slack communication.

Departure from common sense

A straightforward expectation is that making lab knowledge easier to ask about would also make documentation gaps more visible to managers. CHOIR instead surfaces a privacy-awareness tension: people ask privately, so the very mechanism that helps individuals retrieve knowledge can hide missing documentation from directors.

Actual novelty

CHOIR’s novelty is the integrated socio-technical workflow for organizational memory in research labs: document-grounded Q&A, Q&A sharing for follow-up discussion, knowledge extraction from Slack conversations, and AI-assisted document updates. The contribution is not just a chatbot, but a connected workflow that links retrieval, sharing, extraction, and maintenance.

Evidence

The paper reports a one-month deployment of CHOIR in four university research labs with 21 participants. Usage data include 107 questions and 38 document updates, and the discussion highlights privacy-awareness tension plus difficulty generalizing personal experiences into documentation. The evidence supports a field-deployment claim about observed use and emergent tensions, not a causal effectiveness claim.

Limits

Method limits

The study does not provide a controlled baseline or quantitative performance measures beyond descriptive statistics, so it cannot isolate CHOIR’s causal effect relative to alternatives. The evidence is strongest for describing usage patterns, interaction dynamics, and participant-reported tensions.

Deployment limits

The deployment was limited to four university research labs over one month, with a STEM-heavy sample and summer onboarding context. One lab was led by the corresponding author, and the system used prepared seed documents, which may shape behavior and limit transferability.

Boundary conditions

Findings are most applicable to small university research labs that already use chat-based communication and maintain shared documents. The privacy-awareness tension and documentation challenges may be especially salient where contributions are informal, context-specific, and socially sensitive.

Manual check: false

Position in field

This sits at the intersection of CSCW/CHI organizational memory, workplace chat augmentation, and LLM-mediated knowledge management. Its main value is as a field deployment showing how AI support can both improve retrieval and expose tensions around visibility, privacy, and documentation labor.

Abstract

University research labs often rely on chat-based platforms for communication and project management, where valuable knowledge surfaces but is easily lost in message streams. Documentation can preserve knowledge, but it requires ongoing maintenance and is challenging to navigate. Drawing on formative interviews that revealed organizational memory challenges in labs, we designed CHOIR, an LLM-based chatbot that supports organizational memory through four key functions: document-grounded Q&A, Q&A sharing for follow-up discussion, knowledge extraction from conversations, and AI-assisted document updates. We deployed CHOIR in four research labs for one month (n=21), where the lab members asked 107 questions and lab directors updated documents 38 times in the organizational memory. Our findings reveal a privacy-awareness tension: questions were asked privately, limiting directors' visibility into documentation gaps. Students often avoided contribution due to challenges in generalizing personal experiences into universal documentation. We contribute design implications for privacy-preserving awareness and supporting context-specific knowledge documentation.