We sidestep the “silicon psychologist” issue: the LLM simply interprets questions, while all numbers come from structured data. AI explains results, but it can’t rewrite the books.

Sounds like a powerful model if you can get it right