If they can't acknowledge the problem of untrustable results with AI, does "vision" have any relevance beyond keeping executives paid?