Explore how AI models answer difficult workplace questions. Compare their choices, inspect their stated reasons, and see where you would make a different call.
We asked the Index model panel the same yes or no questions. Some answers split the panel; others did not.
See what a finished report looks like.
Questions people ask before they trust a score.
What does the Concord Index measure?
It compares what a company has publicly committed to across 12 value dimensions with how 12 AI models answer the same forced-choice questions. Each report shows agreement and coverage. When several models land too close to separate, it names the tied group and who leads it. A separate set of 10 Hard Questions shows how the models decide, with their stated reasons; that set never scores a company report.
Where does the company evidence come from?
Mostly from the company itself: policies, filings, annual and sustainability reports, and official statements. Founder or CEO interviews count when they speak for the company, and news reporting counts only when at least two independent publishers attribute the same policy to it. Every report lists its sources, and a dimension counts only when an exact quotation supports it.
Why is a company listed without a model match?
Naming a model takes verified evidence on at least 6 of the 12 dimensions. Below that, the report is marked Insufficient public record: it shows the evidence that exists and names no model.
How do I get a report for my company?
Leave your work email and company on any Index page. We review each request and email the report if we run it. Nothing is generated live on the site.