AI Governance & Transparency

How we use AI to surface community insights β€” and what our models can and cannot tell you.

Last updated: April 2026

Important: All AI-generated metrics on this platform β€” including Sentiment Scores, Voter Intent trends, and Support Heatmaps β€” are estimates based on community-submitted data. They do not represent official polling results, scientific surveys, or election forecasts.

How Sentiment Scores Are Calculated

Our Sentiment Score aggregates public posts, poll comments, and forum discussions submitted by registered users on this platform. An AI language model analyzes the tone of this text and assigns a sentiment value between 0 (very negative) and 100 (very positive).

This score reflects community opinion expressed on this platform only β€” it is not a nationally representative sample and is subject to selection bias based on who chooses to participate.

Voter Intent Estimation

Voter Intent trends are derived from voluntary responses to our in-app intent widget. Users self-report their likelihood of voting and their support level for a candidate.

These numbers are not weighted by demographics, geography, or population size. They should be interpreted as a directional signal from engaged platform users, not as a prediction of electoral outcomes.

Geographic Support Heatmaps

Geographic heatmaps display supporter density based on location data voluntarily provided by users during registration or profile setup. Coverage is uneven and does not represent all voters in a given region.

Areas with fewer registered users will appear less active even if real-world support is high.

AI-Generated Content Labeling

Any post on this platform that was drafted using our AI Post Generator is automatically labeled with an ✨ AI-Generated notice visible to all readers. We believe transparency about AI-assisted content is essential for an informed community.

Users may edit AI-generated drafts before publishing. The label indicates the post originated from an AI draft, not that the final content is unedited.

Limitations & Responsible Use
  • AI models can reflect biases present in training data or submitted community content.
  • Sentiment analysis performs less accurately on sarcasm, complex political language, or non-English text.
  • Metrics are updated periodically and may lag real-world developments by hours or days.
  • Do not use these metrics to make legal, financial, or campaign-strategy decisions without consulting qualified experts.
Questions about our AI practices?