Back
Polis: Open Source AI for Large-Scale Open-Ended Feedback
webcompdemocracy.org·compdemocracy.org/polis/
Polis is relevant to AI safety discussions around AI-assisted governance and coordination mechanisms; it offers a concrete example of deploying ML to facilitate large-scale human deliberation, which informs debates about democratic oversight of AI systems.
Metadata
Importance: 45/100tool pagetool
Summary
Polis is an open-source platform using machine learning to enable large-scale, open-ended public deliberation by clustering participant opinions and surfacing areas of consensus. It has been deployed in major civic processes including Taiwan's vTaiwan initiative and Uber regulation discussions. The tool demonstrates how AI can facilitate collective intelligence and reduce polarization in democratic decision-making.
Key Points
- •Uses real-time ML clustering to map opinion landscapes across large groups, identifying consensus and disagreement patterns without predefined answer choices
- •Successfully deployed in high-stakes governance contexts, notably Taiwan's vTaiwan platform for digital policy deliberation
- •Open-source design allows civic technologists and governments to self-host and adapt the platform for local democratic processes
- •Designed to reduce polarization by surfacing statements that bridge across opinion clusters rather than amplifying divisive content
- •Represents a practical application of AI-assisted coordination tools for collective decision-making at scale
Cited by 1 page
| Page | Type | Quality |
|---|---|---|
| AI-Assisted Deliberation | Approach | 63.0 |
Cached Content Preview
HTTP 200Fetched Apr 9, 20264 KB
The Computational Democracy Project | The Computational Democracy Project Polis
In plain language, Polis is a an open source technology for survey research that leverages data science. It is described on the website as:
a real-time system for gathering, analyzing and understanding what large groups of people think in their own words, enabled by advanced statistics and machine learning.
More specifically, Polis is a platform for a conversation , in which participants submit short text statements, or comments , (<140 characters) which are then sent out semi-randomly to other participants to vote on by clicking agree, disagree or pass . Polis allows conversation owners to create conversations which can seamlessly engage (currently) up to hundreds of thousands or (conceivably) millions of participants .
Polis provides automatic (real-time, web-based report ) and CompDem, or your own data scientists, can produce manual (python Jupyter notebooks based) data science analysis of the resulting polis opinion matrix available in the data export . See: services .
Polis is a 💡 Wikisurvey , as:
the dimensions of the survey are created by the participants themselves
the survey adapts to participation over time and makes good use of people's time by showing comments semi-randomly
participants do not need to complete the entire survey to contribute meaning
pol.is is the main instance of the technology hosted online, but there are other instances in the wild.
In its highest ambition, Polis is a platform for enabling collective intelligence in human societies and fostering mutual understanding at scale in the tradition of nonviolent communication . Polis is open source software released under the AGPL3 license
We generally pronounce it like this .
Tom Atlee offers a more holistic narrative description of Polis :
Polis might at first look like a fancy online opinion polling tool. But it can more accurately be described as an evolving, lively, artificially intelligent, interactive “suggestion box” that can realize consensus around a controversial question. It works like this: Dozens or thousands of participants submit short statements expressing their views on the question. They also respond to other participants’ statements with Agree, Disagree, or Pass.
As hundreds of these statements and evaluations accumulate, a behind-the-scenes algorithm sorts participants into diverse clusters of like-minded responders. The algorithm then identifies “consensus statements" about which all the diverse clusters agree. At any time, participants can view graphics revealing the whole ecosystem of perspectives as it evolves. If they wish, they can create new statements that respond to what they see in that overview. Over time their statements tend to become more specific and practical.
After days or weeks of this activity, the results - both consensus statements and divisive statements - are usually turned over to a dialogue among diverse stakeho
... (truncated, 4 KB total)Resource ID:
b7a0eca58a8ad095 | Stable ID: sid_pX0DtdlrVn