And further, if you can't trust the AI to accurately describe itself, how can you trust it on any other topic?