ChatbotLab: Embedding Human-AI Conversations in Research Studies

An open-source backend for integrating LLM-driven chat experiences into surveys and crowdsourced studies

Overview

ChatbotLab lets researchers embed large language model (LLM) conversations directly inside survey platforms such as Qualtrics, REDCap, and LimeSurvey. Researchers can then recruit participants through Prolific or MTurk.

Participants never access ChatbotLab directly. ChatbotLab runs on AWS and serves a secure web interface, loaded within the survey. ChatbotLab stores all conversations and metadata automatically for later analysis.

ChatbotLab requires no programming. Deployment is a single automated step. Every part of a study, including the model, the prompt, and the bot’s behavior, is configured through a web-based admin panel.

Workflow Summary

  1. Deploy ChatbotLab on a cloud-based server (Amazon Web Services), using a single automated workflow

  2. Configure your bot through the admin panel: model, prompt, persona, and conversation behavior

  3. Embed ChatbotLab within Qualtrics, REDCap, or LimeSurvey

  4. Recruit participants through Prolific or MTurk

  5. Collect and export conversation data for analysis

Key Features

  • One-click deployment: automated, containerized hosting on your own AWS account. No server setup or coding is required.

  • Survey integration: embed bots inside Qualtrics, REDCap, or LimeSurvey

  • Crowdsourcing integration: manage participants via Prolific or MTurk

  • Bot configuration: define prompts, personas, models, and delays through a web admin panel

  • Human-like interaction: typing delays, message chunking, personas, and idle follow-ups

  • Data collection: automatically capture conversation and survey metadata, linked per participant

  • Analysis tools: export to ConvoKit, DLATK, or R’s text package for linguistic research