rohinmshah comments on [AN #108]: Why we should scrutinize arguments for AI risk