Human-AI Gap
Where model choices split from the human answer distribution, and where models answer with false certainty on questions people are divided on.
Open Human-AI GapFRI Verticals
The Faith Response Index reports across five verticals: People vs LLMs, Faith vs Secular Choice, Faith-Context Adaptation, Representational Equity, and Model Comparison. Each has its own page and JSON mirror.
Each vertical has a dedicated public page with the headline figures for the current snapshot and a link to the full report in Core. Core stays as the overview that links to all five.
Where model choices split from the human answer distribution, and where models answer with false certainty on questions people are divided on.
Open Human-AI GapHow often models collapse to one option on faith-versus-secular choices, and which direction they collapse toward.
Open Faith/Secular GapWhether models change their practical answer when the user gives clear faith identity, practice, or community context.
Open Faith ContextWhether traditions receive comparable tone, specificity, and respect in controlled representation tasks.
Open Faith EquityThe directional Core FRI score by model, the field median, and the gap from the leader.
Open Model Comparison
Every vertical page has a JSON mirror at /verticals/<slug>.json with the
same headline figures, and an evidence API at /api/v1/verticals/<id>.json
with the full mission-evidence block. Both are public with no sign-in.
Where model choices split from the human answer distribution, and where models answer with false certainty on questions people are divided on.
Open the JSON mirrorHow often models collapse to one option on faith-versus-secular choices, and which direction they collapse toward.
Open the JSON mirrorWhether models change their practical answer when the user gives clear faith identity, practice, or community context.
Open the JSON mirrorWhether traditions receive comparable tone, specificity, and respect in controlled representation tasks.
Open the JSON mirrorThe directional Core FRI score by model, the field median, and the gap from the leader.
Open the JSON mirror