The problem
AI answers sound just as sure when they are wrong as when they are right. As the writing improves, the mistakes get harder to spot.
The idea
Trust should come after a look at the evidence. You should be able to see where an answer came from, and what kind of thing each part is, before you rely on it.
What we’ve built
A working system, not a concept. It runs today on Scripture study: ask a question, and every part of the answer is labelled as source text, scholarship or the machine’s own summary. Those are never blended.
Where it’s going
The same system applied to other fields where a wrong answer costs someone: health, law, finance.
