It's great seeing time and time again that no one really does understand these models and that their preconceived notions of what biases exist ends up shooting them in the foot. It truly shows that they don't really understand how systematically problematic the underlying datasets are and the repurcussions of relying on them too heavily.