Comment by alpha21io

Even if this convergence across models is just an artifact of RLHF or common training data, there's a lot of truth to it. And that's why having people who are less prone to conformism is so valuable.