Neel Nanda comments on There should be a public adversarial collaboration on AI x-risk