Problem Addressed
New volunteers and activists often find the thought of canvassing intimidating with new volunteers often worried about how to have easy and natural conversations. We decided to see if we could use AI to build a bot that would allow campaigners to practise canvassing. This tool started as a prototype idea at one of Campaign Lab’s AI hack days.
Approach & Implementation
The Door Knocking Training Bot is an interactive training platform where campaigners practice doorstep conversations with diverse AI-generated voter personas. Built by veteran Campaign Labbers Peyman and Ben. Hannah created 100 simulated personas, based on the More in Common seven segments using ChatGPT. Each persona had a different reason they would vote Labour and would open the door in a different mood.
The LLM then acts as a different persona and the users then have a doorstep conversation. The conversation is then sent to the LLM to analyse and provide feedback based on a limited sample of 20 “good” doorstep conversations, guides, and training for activists. The bot then gives the user a score out of 10 and lists the strengths and weaknesses of their conversation. The system analyses each conversation against effective canvassing examples, providing personalised feedback, a performance score, and specific areas for improvement.
This tool has now been developed further into a canvass training bot and is related to the Trade Union Rep Training bot – Rep Coach.
Resources
As is the case for all of our tools, iterative improvement is only made possible by testing and user feedback. If you have any feedback for this tool, please feel free to get in touch with: [email protected]