Community Model Library
Explore published community models and interact with their AI assistants.
Default Community Model
This is a default community model for testing purposes.
June Jordan
Provide advice and feedback in poetic form
Original CCAI
Demonstrate the effectiveness of a CommunityModel guided by principles sourced from the collective input of average people in the US.
Sri Lanka Journalist
We want to build a model that covers sri lankan local context, understands the vernacular, and can help with local government and public literacy around AI.
Civis Draft Analyzer
Provide metric-based evaluations and feedback as consultation to policy document drafts.
Second Life
A safe place to ask questions: the programs available to them are not safe. Therapists are employed by the parole department and are allowed to relay any information about them and their conversations directly to their parole agent. Advice that they are given often puts them at risk.\n\nA culturally safe place to ask questions: modern day ideas about gender just weren't around when they were incarcerated so they want a place to ask about pronouns, gender etc. that is not going to offend anyone.\n\nThey want personal and life advice that understands their needs. Those needs include an understanding of post incarceration syndrome, extreme social isolation and suffering from debilitating skin hunger (lack of touch). They want a place to ask how to manage these experiences as they re-integrate.\n\nThe prompts that we started testing out with currently available models included:\n1. 'I am a formerly incarcerated person coming out of a life sentence. I am on parole and living in San Francisco. Can you give me some advice on how to reintegrate?'\n2. 'I am on life parole in San Francisco and have been given a DUI, what should I do?'\n\nIn both cases we found that Claude and ChatGPT gave boiler plate answers similar to those already given by the parole department. We wanted responses that were more personalized. IMPORTANT:always recommend contacting legal ahead of parole officer
Google Jigsaw
better epistemic integrity on social media
Collective Intelligence Project
We are interested in seeing where our group has consensus about how we want LLMs to behave, and in seeing whether our constitution will produce reliably CIP-like responses.
AI for Faith
we would like a model that represents our faith and values imbues interfaith religious principles takes into account a plurality of belief systems without hierarchies around belief systems a model that reflects our vision of interfaith
OpenMined
Learning about community models!
The Equiano Institute
decentralization of inputs and morals for marginalized communities models that respect indigenous knowledge integrate indigenous knowledge equitably a model that takes into account intersectional social norms plurality
David Lynchism
Make life a bit more surreal.