I actually like this. I've dealt with several situation where AI pretends to find a solution or has a false positive because it lacks correct information inside or outside itself. I think this is a fine idea, so long as it's opt-in